Create a useful RPG program to purge unnecessary files from your IFS.
The year-end stuff is over, you've captured all your yearly snapshot data in their own files, and you're feeling pretty good about yourself. Next thing is to clean house. Reports are a good place to start. You likely have a lot of PDFs and Excel spreadsheets that should be purged from your IFS, and this article will walk you through developing a reusable program to easily do exactly that.
Qhell has a find command that we will be using to provide the functionality to delete files that are beyond a certain age. We'll start by discussing the details of the find command interactively in Qhell. Then we'll build a simple program to utilize this command.
Let's jump right in with an example. Then we'll break it apart. Start up Qhell by entering strqsh on your green-screen command line. For our first example, we'll simply list the files in a specific directory that are greater than 90 days old.
find /Public/myDirectory -type f -mtime +90 –print
Here's a breakdown of the command above:
Command Segment |
Description |
Find |
The command that will find the files |
/Public/myDirectory |
The directory to search for the files |
-type f |
Look only at files; no directories will be listed |
-mtime +90 |
Include only files that are greater than 90 days old |
|
Display the results on the screen |
Executing this command will display the list of files that are older than 90 days within the directory path that you specify. Most of the segments are self-explanatory, but the time parameter requires a little more detail to include a few additional time options that may be of interest.
Different Times: mtime, ctime, and atime
There are several ways to specify the age of the files we are looking for, but for this article, I'll be discussing only the times that relate to days. Here's a brief overview of each one that's available:
Time |
Description |
mtime |
Last Modified Time—the last time the data contents were changed |
ctime |
Last Change Time—the last time the attributes of the file were changed |
atime |
Last Access Time—the last time the contents of the file were read |
For this example, we'll be using the last modified time with mtime, which is what I commonly use as my purging criteria. You may want to use another.
In the code above, I used a positive (+) number to find older files that were greater than a specified numbers of days. If I wanted to find newer files that were less than a specified number of days, I would use a negative (-) number.
Filtering Files
Suppose you have a mix of files in your directory of Excel spreadsheets, CSV files, and PDF files with the file name extensions of .xls, .csv, and .pdf, respectively. And you want to select only the Excel spreadsheets. You can do this by filtering the files that are identified using the –name parameter as follows:
find /Public/myDirectory -type f –name '*.xls' -mtime +90 –print
Running this command will list all the files ending with the .xls file name extension. You can be creative with your file pattern recognition, so you're not just limited to file extensions; you could do more.
Preventing Subdirectory Recursion
By default, the find command will automatically search for files that are found in subdirectories of the directory location that you specify. This may or may not be desirable.
For some versions of the find command, there is a –maxdepth option that you can set to the value of 1 to keep the command from going into the subdirectories. Unfortunately, in Qhell, at least in my current version of V7R1, this option is not available.
For UNIX-type commands, there are tons of possibilities to be creative with, but I've found the simplest solution is to utilize the –prune option available with the find command. By specifying your target path to search, you can additionally specify a path to prune, which will eliminate all subdirectories.
find /Public/myDirectory -path '/Public/myDirectory/*' -prune
-type f –name '*.xls' -mtime +90 –print
By using –path, we can specify the directory paths that we want to prevent from being searched, along with the –prune parameter. The –path tells –prune what path to mask from the search, which will be all the subdirectories in this case.
The key difference between the search path and the prune path is the slash/asterisk (/*) character combination, which indicates the subdirectories of the search path.
Deleting Files
Up to this point, we've identified the files by listing them on the screen. This verifies that our filter criteria are working harmlessly with no modification being made to any of the files. Next, we are going to use that information to delete the older files to keep the directories clean.
I always list my files first to make sure that I'm not going to delete more files than I expect to. I suggest that you do the same. Deleting mass files could have disastrous results if you specify the wrong directory. This is why it doesn't hurt to have the *.xls filter to ensure you're deleting only Excel files.
To delete the files, we replace the –print parameter of the find command with the –exec parameter to execute the rm command to remove the files from the IFS, as follows:
find /Public/myDirectory -path '/Public/myDirectory/*' -prune
-type f –name '*.xls' -mtime +90 -exec rm {} \;
The –exec parameter indicates that we want to execute a command, which is followed by the command. The curly braces ({}) will be replaced with the name of each file that was previously listed so that each file listed will have the specified command executed on it. The backslash (\) escapes the command and the semicolon (;) tells the rm command that the list of arguments has ended.
Writing the IFSCLEAN RPG Program
Now that we know how to do everything in Qhell, let's write an RPG program that we can easily reuse to take advantage of the find command to purge our directories.
We will be encapsulating our Qhell code within a procedure named cleanIFS. To support the settings described earlier, our prototype for the procedure will look like this:
D cleanIFS...
D PR 1N
D argPath 2000A const varying
D argDays 3S 0 const
D argRecurse 1N const options(*NOPASS:*OMIT)
D argFilter 128A const options(*NOPASS:*OMIT)
D varying
D argTime 1A const options(*NOPASS:*OMIT)
The parameters will give us the capability to access all of the options that we will want to control for the purging process. Here is the code for the cleanIFS procedure:
*-----------------------------------------------------------------
* cleanIFS: Delete files on IFS using Qhell
*-----------------------------------------------------------------
P cleanIFS...
P B EXPORT
D cleanIFS...
D PI 1N
D argPath 2000A const varying
D argDays 3S 0 const
D argRecurse 1N const options(*NOPASS:*OMIT)
D argFilter 128A const options(*NOPASS:*OMIT)
D varying
D argTime 1A const options(*NOPASS:*OMIT)
D* Local Variables
D svReturn S 1N
D svRecurse S 1N inz(*OFF)
D svFilter S 128A
D svTime S 1A inz('m')
D svCmdString S 512A
/free
svReturn = *OFF;
//---------------------------------------------------------
// Optional Safety Mechanism, only allow full paths in Public
// <Insert Extensive Warning Here for Other Programmers>
//---------------------------------------------------------
if (%subst(argPath:1:8) <> '/Public/');
return *ON;
endif;
As a warning, the Qhell rm command can be very dangerous; if you aren't careful, you could delete system files, which would not be good. So, I try to put in a safety mechanism that will allow only full paths to be specified within the Public folder. I have structured my IFS so that all of my program output goes into subfolders of the Public directory off the root of the IFS. This is not necessary, but I put this out there to make you consciously aware to be careful. Hard-coding a safety mechanism like this forces a programmer to review the code to see what it's doing and consciously add to the permitted list or override it.
To implement this safety net, I am going to ensure that every path that is being purged begins with /Public. If it doesn't, the program will not execute the purge and will pass back a failure value to the calling program.
//----------------------------------------------------------
// Initialize local variable with parameters (if applicable)
//----------------------------------------------------------
if %parms > 2;
if %addr(argRecurse) <> *NULL;
svRecurse = argRecurse;
endif;
endif;
if %parms > 3;
if %addr(argFilter) <> *NULL;
svFilter = %trim(argFilter);
endif;
endif;
if %parms > 4;
if %addr(argTime) <> *NULL;
svTime = argTime;
endif;
endif;
The preceding segment of code is pretty standard for my procedures. Local variables will be defined to act as preprocessed data that will be prepared for usage through the rest of the code. These variables may be passed in through the parameters or set to default values.
The next section of code will start building the string that will be passed to the STRQSH command. The string is initialized with the static value of the command itself, followed by the specified path to purge. The static –type f indicates that we will be processing only files.
//-------------------------------------------------------------
// Start building the Qhell String
svCmdString = 'STRQSH CMD(''find '
+ %trim(argPath)
+ ' -type f ';
Next, we will support the parameters that act as switches to turn certain options on or off. The recursive option will determine if the purge will recursively process all the subdirectories of the primary directory that we are purging.
monitor;
//-------------------------------------------------------------
// Safer Road, Default to Not Recursive
if not svRecurse;
svCmdString = %trim(svCmdString)
+ ' -path '''''
+ %trim(argPath)
+ '/*'''''
+ ' -prune';
endif;
I put a note here that it is safer to prohibit recursion. I say this because, if you have subdirectories, they may have a different purpose than the directory that you are targeting. But that's just a safety tip. If you know you want to delete all files within that directory and all subdirectories, then you would allow recursion to occur. It's pretty common to do this; otherwise, recursion would not be the default. Being specific could help protect you against mistakes. However, this precaution is unnecessary after you've tested out your command interactively. But the capability is still relevant when you want to purge only certain files.
//-------------------------------------------------------------
// Safer Road, Filter on File Name Pattern
if svFilter <> *BLANKS;
svCmdString = %trim(svCmdString)
+ ' -name '''''
+ %trim(svFilter)
+ '''''';
endif;
I also put a note here that it is safer to use a filter. I say this because if you accidentally specified a system directory to purge, changes are you wouldn't find any Excel spreadsheets. And if you did, I wouldn't think they would be critical files anyway. The same thinking applies here that once you've verified that your path is correct, then the safety aspect is unnecessary.
You may notice that I have a lot of single quotes in the string above. This is not a typo. When you want a single quote to be contained in a string without ending it, you double it up to use two quotes to represent a single quote in the results. In this case, the string will be processed twice, once by QCMDEXC and once by the command being executed STRQSH, so we need five total.
The remaining code will set the number of days for the time selection and append the execution command. The command will then be executed within the RPG code:
//-------------------------------------------------------------
// Time: M, C or A
svTime = %xlate('MCA':'mca':svTime);
svCmdString = %trim(svCmdString) + ' -'
+ svTime + 'time';
//-------------------------------------------------------------
// +/- Days
if (argDays >= 0);
svCmdString = %trim(svCmdString) + ' +';
else;
svCmdString = %trim(svCmdString) + ' -';
endif;
svCmdString = %trim(svCmdString)
+ %trim(%editc(%abs(argDays):'J'));
//-------------------------------------------------------------
// -exec rm
svCmdString = %trim(svCmdString)
+ ' -exec rm {} \;'')';
//-------------------------------------------------------------
ExecuteCommand(%trim(svCmdString):%len(%trim(svCmdString)));
on-error;
// Exception
svReturn = *ON;
endmon;
return svReturn;
/end-free
P E
The final part of the service program ensures that the time letter that is passed in is in lowercase and then attaches it to the time parameter of the find command. The days are then formatted to use a sign character as expected by the command. Then the rm execution segment is appended to the end of the command.
Once the string is built for the find command, it is executed with ExecuteCommand, which is just a prototype to QCMDEXC.
To test our new procedure, we will use a simple program to call the new procedure. This program will simply initialize some variables to pass into the parameters of our new cleanIFS procedure. You can take this further by creating a physical file that could contain these settings for different directories. Then you just loop through the directories and pass in the purge settings.
D displayBytes S 52A
D strPath S 512A
D strFilter S 128A
D strTime S 1A
D intDays S 10I 0
D boolRecurse S 1N
D* Prototype for QCMDEXC API
D ExecuteCommand...
D PR extPgm('QCMDEXC')
D argInCommand 65535A const options(*varsize)
D argInLength 15P 5 const
D argInDBCS 3A const options(*nopass)
D*
D* Prototype for cleaning IFS folders
D cleanIFS...
D PR 1N
D argPath 512A const varying
D argDays 3S 0 const
D argRecurse 1N const options(*NOPASS:*OMIT)
D argFilter 128A const options(*NOPASS:*OMIT)
D varying
D argTime 1A const options(*NOPASS:*OMIT)
/free
//-------------------------------------------------------------
// Initialize Parameter Information
strPath = '/Public/myDirectory';
intDays = 90;
boolRecurse = *OFF;
strFilter = '.xls';
strTime = 'm';
//-------------------------------------------------------------
// Display Purging Information
displayBytes = 'Path: ' + %trim(strPath);
dsply displayBytes;
displayBytes = 'Days: ' + %trim(%editc(intDays:'J'))
+ ' (' + strTime + ')time';
dsply displayBytes;
displayBytes = 'Filter: (' + %trim(strFILTER)
+ ') Recursive: ' + boolRecurse;
dsply displayBytes;
//-------------------------------------------------------------
if (cleanIFS(strPath: intDays: boolRecurse: strFilter: strTime));
displayBytes = 'FAIL:( ' + %trim(strPath);
else;
displayBytes = 'PASS:) ' + %trim(strPath);
endif;
dsply displayBytes;
*inlr = *ON;
/end-free
If you run the program, you'll see the following output:
> call mcp049rpg
DSPLY Path: /Public/myDirectory
DSPLY Days: 90 (m)time
DSPLY Filter: (.xls) Recursive: 0
Command ended normally with exit status 0.
DSPLY PASS:) /Public/myDirectory
You will most likely need to change the code to support the directories that you are using for your files. I used the hard-coded /Public/ folder to prevent accidentally purging from an unintended directory. You can either remove or update this setting to match your directory structure.
Take the code for a test drive. You no longer have excuses to postpone those cleanup processes that you've been putting off. I strongly encourage you to run the commands interactively within Qhell that list the files that qualify for purging to become familiar with the commands and ensure that you are deleting the files you intend to purge. Then purge away!
Download the Code
You can download the code used in this article by clicking here.
LATEST COMMENTS
MC Press Online