Windows Batch Files for Fun and Profit

This chapter is from the book

Windows XP Under the Hood: Hardcore Windows Scripting and Command Line Power

Learn More Buy

This chapter is from the book

This chapter is from the book 

Windows XP Under the Hood: Hardcore Windows Scripting and Command Line Power

Learn More Buy

Useful Batch File Techniques

As mentioned in Chapter 10, "Deploying Scripts for Computer and Network Management," I like to make scripts and batch files "user friendly," not only so that they can be used more easily by other people, but also because I know that three weeks after I write a program, I'll have forgotten how to use it, what it does, and how it works. I sometimes find myself standing in front of the refrigerator wondering what I was after when I opened it, so this isn't surprising.

With this in mind, in this section I'll cover a few techniques and tricks that I use to make batch files more helpful, useful, and reliable.

Processing Command-Line Options

You may want your batch files to act like the Windows built-in commands and have them recognize options that begin with / or -. It's very helpful to have a batch file recognize /? as a request to display information about the program itself.

The shift command is useful here. At the beginning of the batch file, you can examine the first argument (%1) and see whether it starts with /. If it does, it's an option. You can set an environment variable, delete the argument with shift, and then repeat the process until all options have been removed.

The following three command options can be useful to implement:

Here's how I process these options at the beginning of a batch file:

@rem                                        Example File batch1204.bat 
@echo off
setlocal
rem - initialize options to their default values
set verbose=NO
set quiet=NO
:again
    set arg=%1
    if "%arg:,1%" == "/" (
        if "/i" %arg% EQU "/v" (
            set verbose=YES
        ) else if "/i" %arg% EQU "/q" (
            set quiet=YES
        ) else (
            :usage
            echo Usage: %0 [/v ^| /q] filename...
            echo.
            echo This batch file copies each of the files to the
            echo network folder. Other helpful information goes here.
            exit /b
        )
        shift /1
        goto again
    )

Here are some notes about the programming in this example:

• To examine the leftmost character of each argument, the program copies it to the environment variable arg and then examines the first character with %arg:,1%. Because the argument could be empty, the first if statement has to use quotes around the strings being compared.
• Because the remaining if commands are only reached if arg starts with /, it definitely is not empty, so there is no need to use quotes around the other tests. However, it's a good habit to use them anyway.
• The final else prints the usage information if the user types /? or any invalid option.
• The shift command uses /1 so that %0 will be left alone and will always contain the name of the batch file.

Later in the batch file, you can use the values of the environment variables verbose and quiet to determine whether to print messages. Extra debugging information can be printed like this:

if "%verbose%" == "YES" echo Got to the part where we'll clean out the folder 

Similarly, normal messages such as Processing file %1 and pause commands that stop the batch file can be disabled if the /Q option is given:

if not "%quiet%" == "YES" echo Processing file %1... 

I use the strange if not test in this case as a failsafe measure: If the variable quiet is not defined, I'd rather have the output appear. (This is an important part of making reliable programs: To the extent possible, have programs anticipate things that should never happen and try to make the best of the situation should they occur.)

Once the batch file has read and eliminated any command-line options, %1 indicates the first filename, or whatever the command-line argument means in your application. If you expect at least one argument, you can print out the usage information and quit if it's missing:

if "%1" == "" goto usage 

This sends the batch program back to the label :usage, which you'll find back in the part of the batch file that processes the command-line options.

Then, if your batch file processes an arbitrary number of input items, you can process each of them in turn using a batch file subroutine:

for each %%x in (%*) do call :process %%x 
exit /b

In this example, %* will be replaced with all the remaining command-line arguments. Any wildcards in the names will be expanded into matching filenames, and the for command will call the batch file subroutine process with each item in turn.

The batch file subroutine process will receive the item as argument number 1, so it will usually start like this:

:process 
if not "%quiet%" == "YES" echo Processing %1...

Managing Network Mappings

If your batch file uses network folders, you may want to map one or more drive letters to shared folders. The problem is that the drive letter you want to use may already be in use. You have to use a creative strategy to deal with this. I have three suggestions.

Use UNC Pathnames

In many cases, you can avoid mapping a drive letter entirely. Most Windows command-line programs will accept UNC pathnames, so you can use the \\server\sharename\... format directly. For example, I used a batch file while I was writing this book to save backup copies of the content on another computer:

@echo off 
echo Backing up chapter files
xcopy c:\book\*.* \\bali\brian\bookbackup /S/Z/Y/M

This copied the files to the server \\bali, to subfolder bookbackup of the shared folder brian.

Use Pushd to Assign a Random Drive Letter

The pushd command will assign a random drive letter if you specify a network path. If you can write your batch file commands to use the current directory so that you don't need to know the drive letter, so much the better. My backup batch file could have been written this way:

pushd \\bali\brian\bookbackup 
xcopy c:\book\*.* . /S/Z/Y/M
popd

You can also determine the drive letter that pushd creates using the %cd% environment variable. This "virtual" variable always returns the current drive letter and path. After the pushd command, it might have the value Y:\.

Be sure to end the batch file with popd to release the temporary drive letter.

Delete Previous Mappings Before Starting

If you want or need to use fixed, mapped network drive letters, you should take a heavy-handed approach to mapping: Delete any preexisting mapping of your desired drive letter first and then map it to the desired network folder.

Because the network drive may or may not be mapped when the batch file starts, I like to use net use to delete any existing mapping. If you redirect its output to the nul file, any error messages simply disappear. This prevents the batch file from displaying an error message if the drive letter is not mapped when it starts. Here's an example:

rem - map drive G to the network folder 

net use g: /del >nul 2>nul        & rem - delete previous mapping if any

net use g: \\server\sharename     & rem - make new mapping
if errorlevel 1 (
    echo The network folder is not available
    if not "%quiet%" == "YES" pause
    exit /b
)

Checking for Correct Arguments

If your batch file accepts filenames as input, it's possible that the batch file's user will type a name incorrectly. It's best to detect this as quickly as possible, rather than letting whatever programs the batch file runs encounter invalid filenames. Before taking any other action with the files, you can perform a quick check of the names the user entered with a loop like this:

for %%f in (%*) do ( 
    if not exist %%f echo Error: file %%f does not exist & exit /b
)

Of course, this assumes that all the command-line arguments are filenames—you will have to write a checking procedure that's appropriate to your own needs. The idea is that whenever possible, you should validate all input before starting to work with it.

Keeping Log Files

If you write batch files to use with the Command Scheduler, you won't be able to see any error messages your batch program prints if it runs into trouble. As mentioned in Chapter 10, it's important to have unattended programs keep a record of their activity so that you can confirm whether they're working correctly; if they're not, you can find out what is going wrong.

When I write a batch file for unattended use, I usually have it create a log file as its first step and store the current time and date in the file. Then, I sprinkle echo commands throughout the program to add a running commentary to the log file. The structure looks like this:

@echo off 
set logfile=MYBATCH.LOG
echo Batch command: %0 %* >%logfile%
echo Started at %date% %time% >>%logfile%
echo ------------------------ >>%logfile%

I use an environment variable for the name of the log file so that if it's necessary to change its name, I only need to edit the set command at the beginning of the file.

Adding the time and date lets you quickly determine whether the scheduled batch process is actually being run.

Then, throughout the batch file you can use echo commands similar to the ones used at startup to add to the log file. Record the names of files processed, add a note at the beginning of each major group of commands, and definitely record any problems encountered. It's best to display messages to the standard output as well as to the log file:

@rem                                        Example File batch1205.bat 
set logfile = mappit.log

echo Mapping network drive...
echo Mapping network drive... >>%logfile%

net use m: \\server\sharename
if errorlevel 1 (
    echo Unable to use shared folder \\server\sharename
    echo Unable to use shared folder \\server\sharename >>%logfile%
    echo Quitting prematurely! >>%logfile%
    exit /b
)

This way, you can see the output yourself if you run the batch file manually.

Finally, it's a good idea to record a final entry just before the batch file exits so that you know it didn't get stuck just before the last command. The final commands in the batch file's main program section might look like this:

echo ---------------------- >>%logfile% 
echo Ended at %date% %time% >>%logfile%

In fact, you might want to have the program jump down to these final statements even after encountering a fatal error. Instead of using exit /b after encountering the network error, I could have written goto done and put this at the end of the batch file:

:done 
echo ---------------------- >>%logfile%
echo Ended at %date% %time% >>%logfile%

This way, the log file should always end like this:

------------------------ 
Ended at Thu 05/02/2002 23:54:12.90

If it doesn't, I'll know that something quite unexpected must have happened.