Speaking UNIX, Part 1: Command the power of the command line

Mix and match UNIX utilities to create impromptu programs

Learn the basics of the UNIX shell and discover how you can use the command line to combine the finite set of UNIX utilities into innumerable data transforms.

Share:

Martin Streicher (martin.streicher@gmail.com), Chief Technology Officer, McClatchy Interactive

Photo of Martin StreicherMartin Streicher is a freelance Ruby on Rails developer and the former Editor-in-Chief of Linux Magazine. Martin holds a Masters of Science degree in computer science from Purdue University and has programmed UNIX-like systems since 1986. He collects art and toys. You can reach Martin at martin.streicher@gmail.com.



07 March 2006

Also available in Russian

Speaking UNIX: Hello, shell

One of the most novel and differentiating features of a UNIX® system is its command line. With just a few keystrokes, including a bit of "glue", you can use the command line to combine the finite set of UNIX utilities into innumerable, impromptu data transforms.

For example, to find the list of unique filenames in the folder hierarchy rooted at the current working directory, you can type the following at your shell prompt:

find . -type f -print | sort | uniq

This command line combines three separate utilities:

  • find plumbs the depths of the named directory -- in this case, the file system starting at . or dot (shorthand for the current working directory) -- and emits the names of all entries that match the given criteria. Here, -type f directs find to discover only plain files.

  • sort, as its name implies, processes a list and emits a new list that's sorted alphabetically.

  • uniq (pronounced "unique") scans a list, comparing adjacent elements in the list and removing any duplicates. For instance, suppose you have this list:

    Listing 1. Example list
                            Groucho
    Groucho
    Chico
    Chico
    Groucho
    Harpo
    Zeppo
    Zeppo

    uniq reduces the list to the following:

    Listing 2. uniq command
                            Groucho
    Chico
    Groucho
    Harpo
    Zeppo

    However, if the original list of Marx Brothers is sorted first (reordering all occurrences of a name into a continuous run), running uniq yields this result:

    Listing 3. Running uniq
                            Chico
    Groucho
    Harpo
    Zeppo

To learn more about the extensive features of find, sort, and uniq, refer to each utility's man page on your UNIX system.


Data in, data out, data all about

Used independently, find always takes the contents of the file system as its input data. However, both sort and uniq require data entry or input from the standard input device (stdin). Most often, you provide stdin using the keyboard: You type the data you want sorted on a series of lines, for example.

By default, find prints results on the standard output device (stdout), which is usually your terminal window. Both sort and uniq print outcomes to stdout.

To demonstrate stdin and stdout, type the following text in your terminal window (assume that the leading percent sign (%) is your shell prompt):

Listing 4. stdin and stdout
                % sort
mustache
horn
hat
Control-D

sort reads the three lines you typed from stdin, sorts them, and writes the result to stdout. Figure 1 presents a conceptual picture of running sort, and most UNIX command-line utilities, from the command line.

Figure 1. A typical UNIX command-line utility reads from stdin and writes to stdout
UNIX command-line utility

Some utilities, such as find, don't read from stdin. Instead, they read the data they should process from system resources, such as the file system or the system kernel, and write results to stdout. To visualize how find works, look at Figure 2 below.

Figure 2. Some utilities read data from system resources and write results to stdout
Some utilities read data from system resources and write results to stdout

In addition to using stdin and stdout, UNIX commands can emit error messages to a special outlet that's set aside (by convention, not mandate) for diagnostics. The outlet is called the standard error device (usually referred to as stderr). Figure 3 illustrates a simple command line running a utility.

Figure 3. UNIX commands emit errors to a special channel, standard error
UNIX commands emit errors to a special channel, standard error

As shown in Figure 3, most UNIX commands read input from the terminal, send results to the terminal, and print errors to the terminal. By default, and unless you specify otherwise, your terminal is the source of data for stdin and the destination for both stdout and stderr.


Routing traffic to and fro

However, you can change the source of stdin and the eventual destinations of both stdout and stderr. You can force stdin to read from a text file, a device (say, a probe connected to the computer), or a network connection. Comparably, you can send output to a file, a device, or aconnection. In UNIX, where everything is a file, one source or destination is just as easy to consume or produce as another.

Changing the source and destination of a process's data is referred to as redirection. You redirect stdin to read data from a file or other source; and you can redirect stdout and stderr (separately) to write data somewhere other than the terminal window. In many cases, as in the original find command shown earlier, you can also redirect utilities to consume and produce data from and for other utilities. That is the purpose of the pipe (|). In a command, you can daisy-chain processes together using pipes, sending the data of one command to the next command, just like segments of copper pipe route water from your water heater to your sink.

Figure 4 shows a conceptualization of the find . -type f -print | sort | uniq command.

Figure 4. A conceptual model of three utilities linked by pipes
A conceptual model of three utilities linked by pipes

The stdout of find becomes the stdin of uniq; in turn, the stdout of uniq becomes the stdin for sort. Finally, sort prints the results to its standard output device, which remains connected to the terminal. The stderr of the commands wasn't redirected, so all three utilities print error messages to the terminal. (Error messages from the three utilities are intermixed, but the order of the messages will be correct.)

If necessary, you can extend the pipeline shown above further and redirect the output of uniq to yet another utility. Just tack on another pipe to extend the transform further. For instance, you can append | less to paginate the output using less, or you can add | wc -l to find the number of unique filenames. (wc is an acronym for word count; wc can count characters, words, and lines.)

Alternatively, you can use > to save the output of the entire sequence to a file (destroying the existing contents of the file, if any). You can also use >> to append the results to an existing file (creating the file if it doesn't exist).

Another helpful redirection is <. Figure 5 shows how stdin can be redirected to read from a file. The command sort reads a list of words from the named file and alphabetizes them.

Figure 5. Redirecting standard input to read from a file
Redirecting standard input to read from a file

Often, you'll want to capture stdout and stderr. For example, if you're running a large data-mining task, you might want to review the interim output and any errors that occurred during execution. You can use variants of the redirection syntax to do just that: |&, >&, >>&, pipe, create, and append stdout and stderr simultaneously, respectively. Figure 6 illustrates how stdout and stderr are combined into one output stream.

Figure 6. Combining the standard output and standard error devices
Combining the standard output and standard error devices

Introducing the Z shell

Most modern UNIX shells -- including the Bourne shell (bash) and the Korn shell (ksh) -- support the redirections mentioned here, although the specific syntax required by both of those shells might differ slightly. (Check your shell's documentation for specifics.).

Most of the redirection operators have been consistent features of all UNIX shells for at least 25 years. However, most of those shells have failed to break new ground and explore new ways to apply redirection. For instance, most shells can only redirect input from a single file, and you must use a utility like tee to output to more than one destination. (Like the tee junction used by plumbers, tee has one input and two outputs.) Here's an example using bash as the shell (the command-line interpreter):

Listing 5. bash example
bash$ ls
tellme
bash$ cat tellme
echo Your current login, working directory, and system are...
whoami
pwd
systemname
bash$ bash < tellme |& tee log
Your current login and working directory are...
strike
/home/strike
bash: systemname: command not found
bash$ ls
tellme log
bash$ cat log
Your current login and working directory are
strike
/home/strike
bash: systemname: command not found

Although UNIX shells are highly specialized and generally used interactively using the keyboard, a shell such as bash can also read input from a file. (After all, stdin is just a file.) In the previous snippet, the phrase bash < commands makes bash execute a list of commands found in the file tellme. The phrase |&tee log pipes the stdout and stderr of bash to the tee utility, which prints its stdin to stdout and to the file log.

But what if you want bash to process more than one input file? cat file1 file2 file3 | bash is a workable solution, and perhaps the only one, because bash doesn't support syntax like bash < file1 < file2 < file3.

Moreover, bash can't redirect output to more than one destination. For example, you can't enter an instruction like bighairyscript > ~/log | mail -s "Important stuff" team from the bash command line.

But a relatively new shell, the Z shell (zsh; see Resources), can process multiple input and output redirections within the same command line. For example, here's a command that saves stdout in a file called log and sends it to you using e-mail:

Listing 6. Z shell
zsh% bash < tellme > log | mail -s "Who you are" 'whoami'
bash: line 4: systemname: command not found
zsh% <log
Your current login, working directory, and system are...
strike
/home/strike

(The phrase 'whoami' runs the command whoami and inserts the result of that command in place of the phrase. It's like running a little shell command before the rest of the command line runs.)

Let's walk through the previous command from left to right. The bash command creates the file log and mails the stdout of the commands found in tellme to you. Because stderr wasn't redirected by the > or the pipe, error messages are printed to stdout. The command <log is another Z shell shortcut; it's the same as cat. (And yes, the command > file is equivalent to cat > file.)

The Z shell can also process more than one input redirection. The Z shell command line cat < file1 < file2 < file3 is the same as cat file1 file2 file3. Admittedly, the former syntax is more unwieldy than the latter and, in general, multiple stdout redirection is used far more often. However, if the utility you want to run doesn't accept multiple input arguments, the Z shell's multiple input redirection can come in handy.

The Z shell is full of other novel tricks, including better globbing (wildcard matches), advanced pattern matches, and an extensive automatic completion system that minimizes what you have to type at the command line. The next two articles in this series will delve further into the Z shell.


Shell tricks

Here are some powerful command-line combinations that are sure to make you more productive. The commands should work in all shells, not just zsh.

  • Create a verbatim copy of any directory, including symbolic links, with tar:

    tar cf - /path/to/original |  \
      (mkdir -p /path/to/copy; cd /path/to/copy; tar xvf -)

    The first tar archives the directory /path/to/original and emits the archive file to stdout; the hyphen (-) used with the create (c) option specifies stdout. The command in parentheses is a subshell: Commands in the subshell don't affect the environment of the current shell. mkdir -p creates the named directory, including any intermediate directories that need to be created; and cd changes to the new directory. The second tar reads an archive from stdin and expands it in place; the hyphen used with the extract (x) option refers to stdin.

  • To save the stdout of a command sequence and view it at the same time, use less -O file. The -O option copies stdin to the named file. Here's an example:

    sort /etc/aliases | less -Osorted
  • If a directory contains thousands of files, your shell (including zsh, depending on the number of files and their names) might not be able to enumerate all the files using wildcard matches, because the command line is typically limited to a certain number of characters. Hence, shell script phrases such as

    foreach i (*)
    ...
    end

    might fail. (You'll probably see a message like Line length exceeded when you exceed the length of your command line.) If such an error occurs, use a pipe and the xargs utility. xargs reads data from a pipe and runs a specified command for every line read.

    For instance, if you want to find all Web pages on your server that reference www.example.com, you can use this command line:

    % find / -name '*html' -print \
      | xargs grep -l 'www.example.com' \
      | less -Opages
    

    xargs consumes the filenames from find and runs grep -l repeatedly to process every file, no matter how many files are named. (grep -l prints the name of the file if a match is found and then stops further matching in that file.) less allows you to page through the results and saves the list in the file named pages. The result is a list of filenames that contain the string "www.example.com".


The journey begins

In this article, you learned the basics of the UNIX shell. Subsequent articles will delve more deeply into the plethora of command line tools and techniques at your disposal. From file systems to entire local area networks, virtually all information and systems administration can be undertaken efficiently from the UNIX command line.

Stay tuned!

Resources

Learn

Get products and technologies

Discuss

Comments

developerWorks: Sign in

Required fields are indicated with an asterisk (*).


Need an IBM ID?
Forgot your IBM ID?


Forgot your password?
Change your password

By clicking Submit, you agree to the developerWorks terms of use.

 


The first time you sign into developerWorks, a profile is created for you. Information in your profile (your name, country/region, and company name) is displayed to the public and will accompany any content you post, unless you opt to hide your company name. You may update your IBM account at any time.

All information submitted is secure.

Choose your display name



The first time you sign in to developerWorks, a profile is created for you, so you need to choose a display name. Your display name accompanies the content you post on developerWorks.

Please choose a display name between 3-31 characters. Your display name must be unique in the developerWorks community and should not be your email address for privacy reasons.

Required fields are indicated with an asterisk (*).

(Must be between 3 – 31 characters.)

By clicking Submit, you agree to the developerWorks terms of use.

 


All information submitted is secure.

Dig deeper into AIX and Unix on developerWorks


static.content.url=http://www.ibm.com/developerworks/js/artrating/
SITE_ID=1
Zone=AIX and UNIX
ArticleID=105078
ArticleTitle=Speaking UNIX, Part 1: Command the power of the command line
publish-date=03072006