Batch Script — Variables

There are two types of variables in batch files. One is for parameters which can be passed when the batch file is called and the other is done via the set command.

Command Line Arguments

Batch scripts support the concept of command line arguments wherein arguments can be passed to the batch file when invoked. The arguments can be called from the batch files through the variables %1, %2, %3, and so on.

The following example shows a batch file which accepts 3 command line arguments and echo’s them to the command line screen.

If the above batch script is stored in a file called test.bat and we were to run the batch as

Following is a screenshot of how this would look in the command prompt when the batch file is executed.

The above command produces the following output.

If we were to run the batch as

The output would still remain the same as above. However, the fourth parameter would be ignored.

Set Command

The other way in which variables can be initialized is via the ‘set’ command. Following is the syntax of the set command.

Syntax

variable-name is the name of the variable you want to set.

value is the value which needs to be set against the variable.

/A – This switch is used if the value needs to be numeric in nature.

The following example shows a simple way the set command can be used.

Example

In the above code snippet, a variable called message is defined and set with the value of «Hello World».

To display the value of the variable, note that the variable needs to be enclosed in the % sign.

Output

The above command produces the following output.

Working with Numeric Values

In batch script, it is also possible to define a variable to hold a numeric value. This can be done by using the /A switch.

The following code shows a simple way in which numeric values can be set with the /A switch.

We are first setting the value of 2 variables, a and b to 5 and 10 respectively.

We are adding those values and storing in the variable c.

Finally, we are displaying the value of the variable c.

The output of the above program would be 15.

All of the arithmetic operators work in batch files. The following example shows arithmetic operators can be used in batch files.

The above command produces the following output.

Local vs Global Variables

In any programming language, there is an option to mark variables as having some sort of scope, i.e. the section of code on which they can be accessed. Normally, variable having a global scope can be accessed anywhere from a program whereas local scoped variables have a defined boundary in which they can be accessed.

DOS scripting also has a definition for locally and globally scoped variables. By default, variables are global to your entire command prompt session. Call the SETLOCAL command to make variables local to the scope of your script. After calling SETLOCAL, any variable assignments revert upon calling ENDLOCAL, calling EXIT, or when execution reaches the end of file (EOF) in your script. The following example shows the difference when local and global variables are set in the script.

Example

Few key things to note about the above program.

The ‘globalvar’ is defined with a global scope and is available throughout the entire script.

The ‘var‘ variable is defined in a local scope because it is enclosed between a ‘SETLOCAL’ and ‘ENDLOCAL’ block. Hence, this variable will be destroyed as soon the ‘ENDLOCAL’ statement is executed.

Output

The above command produces the following output.

You will notice that the command echo %var% will not yield anything because after the ENDLOCAL statement, the ‘var’ variable will no longer exist.

Working with Environment Variables

If you have variables that would be used across batch files, then it is always preferable to use environment variables. Once the environment variable is defined, it can be accessed via the % sign. The following example shows how to see the JAVA_HOME defined on a system. The JAVA_HOME variable is a key component that is normally used by a wide variety of applications.

The output would show the JAVA_HOME directory which would depend from system to system. Following is an example of an output.

/* steve jansen */

// another day in paradise hacking code and more

Windows Batch Scripting: Variables

Today we’ll cover variables, which are going to be necessary in any non-trivial batch programs. The syntax for variables can be a bit odd, so it will help to be able to understand a variable and how it’s being used.

Variable Declaration

DOS does not require declaration of variables. The value of undeclared/uninitialized variables is an empty string, or «» . Most people like this, as it reduces the amount of code to write. Personally, I’d like the option to require a variable is declared before it’s used, as this catches silly bugs like typos in variable names.

Variable Assignment

The SET command assigns a value to a variable.

NOTE: Do not use whitespace between the name and value; SET foo = bar will not work but SET foo=bar will work.

The /A switch supports arthimetic operations during assigments. This is a useful tool if you need to validated that user input is a numerical value.

A common convention is to use lowercase names for your script’s variables. System-wide variables, known as environmental variables, use uppercase names. These environmental describe where to find certain things in your system, such as %TEMP% which is path for temporary files. DOS is case insensitive, so this convention isn’t enforced but it’s a good idea to make your script’s easier to read and troubleshoot.

WARNING: SET will always overwrite (clobber) any existing variables. It’s a good idea to verify you aren’t overwriting a system-wide variable when writing a script. A quick ECHO %foo% will confirm that the variable foo isn’t an existing variable. For example, it might be tempting to name a variable “temp”, but, that would change the meaning of the widely used “%TEMP%” environmental varible. DOS includes some “dynamic” environmental variables that behave more like commands. These dynamic varibles include %DATE% , %RANDOM% , and %CD% . It would be a bad idea to overwrite these dynamic variables.

Reading the Value of a Variable

In most situations you can read the value of a variable by prefixing and postfixing the variable name with the % operator. The example below prints the current value of the variable foo to the console output.

There are some special situations in which variables do not use this % syntax. We’ll discuss these special cases later in this series.

Listing Existing Variables

The SET command with no arguments will list all variables for the current command prompt session. Most of these varaiables will be system-wide environmental variables, like %PATH% or %TEMP% .

NOTE: Calling SET will list all regular (static) variables for the current session. This listing excludes the dynamic environmental variables like %DATE% or %CD% . You can list these dynamic variables by viewing the end of the help text for SET, invoked by calling SET /?

Variable Scope (Global vs Local)

By default, variables are global to your entire command prompt session. Call the SETLOCAL command to make variables local to the scope of your script. After calling SETLOCAL , any variable assignments revert upon calling ENDLOCAL , calling EXIT , or when execution reaches the end of file (EOF) in your script.

This example demonstrates changing an existing variable named foo within a script named HelloWorld.cmd . The shell restores the original value of %foo% when HelloWorld.cmd exits.

A real life example might be a script that modifies the system-wide %PATH% environmental variable, which is the list of directories to search for a command when executing a command.

Special Variables

There are a few special situations where variables work a bit differently. The arguments passed on the command line to your script are also variables, but, don’t use the %var% syntax. Rather, you read each argument using a single % with a digit 0-9, representing the ordinal position of the argument. You’ll see this same style used later with a hack to create functions/subroutines in batch scripts.

There is also a variable syntax using ! , like !var! . This is a special type of situation called delayed expansion. You’ll learn more about delayed expansion in when we discuss conditionals (if/then) and looping.

Command Line Arguments to Your Script

You can read the command line arguments passed to your script using a special syntax. The syntax is a single % character followed by the ordinal position of the argument from 0 – 9 . The zero ordinal argument is the name of the batch file itself. So the variable %0 in our script HelloWorld.cmd will be “HelloWorld.cmd”.

The command line argument variables are * %0 : the name of the script/program as called on the command line; always a non-empty value * %1 : the first command line argument; empty if no arguments were provided * %2 : the second command line argument; empty if a second argument wasn’t provided * …: * %9 : the ninth command line argument

NOTE: DOS does support more than 9 command line arguments, however, you cannot directly read the 10th argument of higher. This is because the special variable syntax doesn’t recognize %10 or higher. In fact, the shell reads %10 as postfix the %0 command line argument with the string “0”. Use the SHIFT command to pop the first argument from the list of arguments, which “shifts” all arguments one place to the left. For example, the the second argument shifts from position %2 to %1 , which then exposes the 10th argument as %9 . You will learn how to process a large number of arguments in a loop later in this series.

Tricks with Command Line Arguments

Command Line Arguments also support some really useful optional syntax to run quasi-macros on command line arguments that are file paths. These macros are called variable substitution support and can resolve the path, timestamp, or size of file that is a command line argument. The documentation for this super useful feature is a bit hard to find – run ‘FOR /?’ and page to the end of the output.

I removes quotes from the first command line argument, which is super useful when working with arguments to file paths. You will need to quote any file paths, but, quoting a file path twice will cause a file not found error.

fI is the full path to the folder of the first command line argument

fsI is the same as above but the extra s option yields the DOS 8.3 short name path to the first command line argument (e.g., C:\PROGRA

1 is usually the 8.3 short name variant of C:\Program Files ). This can be helpful when using third party scripts or programs that don’t handle spaces in file paths.

dpI is the full path to the parent folder of the first command line argument. I use this trick in nearly every batch file I write to determine where the script file itself lives. The syntax SET parent=%

dp0 will put the path of the folder for the script file in the variable %parent% .

nxI is just the file name and file extension of the first command line argument. I also use this trick frequently to determine the name of the script at runtime. If I need to print messages to the user, I like to prefix the message with the script’s name, like ECHO %

n0: some message instead of ECHO some message . The prefixing helps the end user by knowing the output is from the script and not another program being called by the script. It may sound silly until you spend hours trying to track down an obtuse error message generated by a script. This is a nice piece of polish I picked up from the Unix/Linux world.

Some Final Polish

I always include these commands at the top of my batch scripts:

The SETLOCAL command ensures that I don’t clobber any existing variables after my script exits. The ENABLEEXTENSIONS argument turns on a very helpful feature called command processor extensions. Trust me, you want command processor extensions. I also store the name of the script (without the file extension) in a variable named %me% ; I use this variable as the prefix to any printed messages (e.g. ECHO %me%: some message ). I also store the parent path to the script in a variable named %parent% . I use this variable to make fully qualified filepaths to any other files in the same directory as our script.

Posted by Steve Jansen Mar 1 st , 2013 batch, scripting, shell, windows

Comments

Hi, I’m Steve. I’m a software developer loving life in Charlotte, NC, an (ISC) 2 CSSLP and an avid fan of Crossfit.

And, no, I’m not Steve Jansen the British jazz drummer, though that does sound like a sweet career.

Guides

Recent Posts

Social Stuff

• @steve-jansen on GitHub
• @steve-jansen on StackOverflow
• @steve-jansen ProTips on Coderwall
• @steve-jansen on Microsft Connect
• @steve-jansen on ASP.NET User Voice
• Subscribe via RSS

Copyright © 2015 — Steve Jansen — Powered by Octopress

How to substitute variable contents in a Windows batch file

I’m writing a simple script to substitute text in an environment variable with other text. The trouble I get is with the substituted or substituted text being pulled from other variables

Works fine (output is ‘The fat cat’ and ‘The thin cat’

However, if ‘fat’ or ‘thin’ are in variables, it doesn’t work

I know this can be done as I’ve seen it in blogs before, but I never saved the link to the blogs.

Anyone have any ideas?

PS. I’m running the scripts on Windows 7

PPS. I know this is easier in Perl/Python/other script language of choice, but I just want to know why something that should be easy is not immediately obvious.

PPPS. I’ve also tried the scripts with delayed expansion explicitly turned on

This makes no difference.

7 Answers 7

Please try the following:

Copy and paste the code into Notepad and save it as a batch file.

I hope you’re convinced!

Use CALL. Put the following in a batch script and run it:

As stated here, we use the fact that:

internal_cmd Run an internal command, first expanding any variables in the argument.

In our case internal_cmd is initially set a=%%a:%b%^=%c%%%.

After expansion internal_cmd becomes set a=%a:fat=thin%.

Thus, in our case running

is equivalent to running:

The problem with:

is that it tries to expand two variables: a: and =thin with a c constant string between them.

The first command outputs:

which is piped into a second command shell for evaluation.

And. How about this?

Recently I came accross the same situation..As said earlier, I used like below and worked.

I try to avoid using SETLOCAL enabledelayedexpansion . ENDLOCAL all (or nearly all) the time, because usually I want to set or modify a few variables and I want the new values to be available in other areas of the script or after the batch script ends (SETLOCAL|ENDLOCAL will forget about any new variables or changes to variables in the «SETLOCAL» part of the script. Sometimes that’s handy, but for me I find it’s usually not.

Currently I use the method described by @Zuzel, but before I knew about that method, I used to use this, which is very similar (but looks a bit more complicated):

the output from running the script:

I like this method because you can call external programs (or internal commands) using modified variables and also capture and process the output of the command (line by line).

But, Zuzel’s method is simpler and cleaner for most situations, including the one you described.

Note:

Both of these methods (and also SETLOCAL enabledelayedexpansion . ENDLOCAL , of course), only work correctly if run from within a batch script. If you try to use either of these two methods («call» or «for») directly in a command prompt window, you will get something different from what the output was running from a script.

For example, run this as a script:

the output from running the script:

Now, run those commands directly in a command prompt window:

examine the before and after output lines from the command prompt window:

Notice that the substitutions are made correctly, but also notice that with running these commands directly in the command prompt window, it adds a set of «%» (percent signs) before and after the expected value each time the substitution is made. So, it makes it difficult to test any of these methods directly in the command prompt window.