PBJ 1.0.4 – New password options and start of CLI

The PBJ library has been getting a lot of attention lately from various developers using it to integrate with their own software and projects. Francisco Amores did a great blog post about using PBJ to help with data loading in an FDMEE project. Probably the coolest thing about his efforts is that it’s  use-case I never imagined: using PBJ in Jython to access PBCS.

One of the things that has been so great about collaborating with Francisco is getting targeted, useful, and practical comments on how he’s using the library and how it can be made better. And I have found time to make various improvements, enhancements, and fix bugs to address his feedback. This is one of the greatest things about open source software.

Continue Reading…

PBJ (PBCS REST API Library for Java) Updates

I have recently gotten quite a bit of feedback from people using the PBJ library to consume and work with the PBCS REST API. This has resulted in a few fixes and improvements. First of all, file uploads and downloads are now working. The code has been in for awhile but wasn’t merged to the master branch.

Second, I have been working with the very talented Francisco Amores to help integrate the PBJ library with FDMEE so that it can be used there in various integration scenarios. So this is a really cool usage of PBJ where the library is being dropped in to FDMEE/ODI and utilized with a very simple Jython script.

To help with this scenario, I added a new compile option to PBJ that allows it to be packaged up into a single “uber JAR” – meaning a Java JAR file that contains all of its dependencies rolled into a single library. This makes it a little easier to integrate and drop-in to other systems, instead of having to worry out additional JAR files.

We had to make a couple of other tweaks to the way the library is packaged in order to make it specifically work in FDMEE/ODI, due to a conflicting underlying library. This is kind of a classic Java class loader problem, because what happens is that the two different versions of the class are both available to be loaded, and the older version of the class gets loaded but that class doesn’t have a required method, so a “no such method found” exception is thrown. But by renaming the package/method when it’s compiled, we can get around it and make it really simple. The PBJ GitHub page has some more info on how to compile this.

I think in the near future Fransisco will be blogging out this really interesting integration scenario, so stay tuned!

Code This, Not That: Python date string formatting

Continuing on in the Code This, Not That series, I’m going to point out some convenient functions for formatting dates. Since Jython runs in the JVM, we get easy access to all of the libraries that the rich Java programming language provides.

Working with FDMEE and custom scripting, it’s likely that you’re going to need to do some work with dates. This will often involve the Java “Date” object. You might find that you need to do something with the date – get the year, get the name of the month, create a string based on the day of the month, and so on. First off, there are a few things you definitely don’t want to do.

One, you don’t want to convert the date object to a string and then start doing string manipulation on them to get the data you want. If you are converting some VBScript code you might be tempted to do something like Year = RIGHT(someDate, 4). This can work but it’s a bit of a hack. What you want to do is work with the date object directly and get some other tried and true code to do the heavy lifting for you.

There are other ways to achieve this, but a pretty simple and robust one is the SimpleDateFormat class. The SimpleDateFormat class allows us to specify a formatting code, then supply it a Date object, and it’ll generate a string for us.

For example, the formatting code for the abbreviation for the month of a date object is “MMM”. Therefore, we can create a SimpleDateFormat object with this code, pass it a date, and the resulting string is Mar or Apr or whatever the current month is:

import java.text.SimpleDateFormat as Sdf
import java.lang.System as System
import java.sql.Date as Date

date = Date(System.currentTimeMillis())

monthShortName = Sdf('MMM')
print monthShortName.format(date)
# prints "Mar" (or whatever the current month is)

There are plenty of other codes that take care of formatting the minutes, hours, seconds, year, and more for us. We can even combine codes in a single string, like the following:

dateFolder = Sdf('yyyyMMdd')
print dateFolder.format(date)

In this case, the format is the four digit year, followed immediately by the two digit month (including a leading zero if it’s less than 10), and the two digit day. You can refer to Oracle’s documentation for more codes over here. You can also put in other characters that might be ignored in the formatting string, such as hyphens between codes, colons, spaces, and more.

Also keep in mind that you create the formatting object and use it as many times as you want to format dates into strings. In other words, if for some reason you need to format many dates into strings using the same format, you just have to create one SimpleDateFormat object and then just use it over and over again.

As I mentioned, there are plenty of ways to go about formatting dates in Java (Jython) but this is a particularly flexible and useful approach. With just a few lines of code you can generate all sorts of formats that you might need. Consider this an FDMEE/Jython pattern – and anti-patterns to this might include the following code situations you can cleanup and make more readable:

  • Treating dates like strings and extracting data via substrings
  • Prefixing “20” on a two-digit year to get a four digit year
  • Keeping a table of month names in an array and using the month number to get the proper month name
  • Keeping a table of month names and then taking the first three characters to get the “short” month name

Lastly, note that the fdmContext["PERIODKEY"] object (which is where you are likely to be doing some date processing) comes in as a java Date object (in other words, if you are treating it as a string object, you are actually relying on the Jython interpreter silently casting it to a string for you, when you could just be working with it directly). For example, to get the numeric four digit year from the PERIODKEY object, you could do this:

# Note the explicit conversion to an integer
year = int(Sdf('yyyy').format(fdmContext["PERIODKEY"]))

Happy coding!

Code This, Not That: From VBScript to Jython

I recently completed a fairly complex VBScript to Jython import integration script conversion. For those that are unfamiliar, Jython is the preferred scripting language of FDMEE (the successor to FDM). Jython offers numerous advantages over VBScript. Chief among those is that the language is more expressive and offers a lot of syntactic sugar, particularly compared to VBScript.

People that know me know that I have a passion for programming. I love writing code in many languages – Java is by far my strongest language, but I also am quite fond of Objective-C (with Swift on my list of things to play with), Python, and a few others. That said I don’t just like to write good, working code – I challenge myself to write clean, elegant, high-performance, easily maintainable, easy to understand, deceptively simple, and fluid code that reads like prose.

One of the greatest compliments I ever got from someone was on an Essbase automation system I had designed. The compliment was, “You made it look simple. And I know it’s not simple – it’s just that good.” I was beaming for days.

But what does this all have with FDMEE and Jython?

Think about this: code is read many more times than it is written. Therefore, it behooves us to write it with as much expressiveness and conciseness as we can, to add useful comments, to be idiomatic with respect to the language we are writing in. Along these lines, in my recent conversion project I was reminded of all the opportunities for writing better code – more expressive code – that Jython (Python) affords us over VBScript.  So to point out some of those examples, over the coming weeks I thought I would write about come particular code examples between VBScript and Jython that I think are worth pointing out, along with some general code recommendations I might have.

So let’s kick things off.

Checking if a value is one of a list of values

It’s common in an import integration script to check if some value, say, the current member from the Accounts dimension, is a particular account. The most straightforward way to handle this condition is with the following code:

if account == '0170100' or account == '0170200' or account == '0170300':
    print "Matched account!"

Thinking back to the notion of reading code more times than we read it, the English language interpretation of this code might be “If the account is 0170100, or if the account is 0170200, or if the account is 0170300, then do this thing.” You might call this the “brute force” approach – to just literally test each condition we might handle.

Can we come up with something a little more… refined? Python offers a convenient syntax that can make this code a little more readable. In Python we can use tuples. A tuple is like a variable that is a bag of other values or variables. Think of it as a simple collection. Using the “in” operator, we can test for membership in a tuple. The following code works the exact same way as the above code:

if account in ('0170100', '0170200', '0170300'):
    print "Matched account using tuple!"

Now the English interpretation of the code might be “Is the account one of these accounts?” The code is shorter, simpler, and reads a little more easily. While this example is somewhat trivial, it’s a good example of leveraging this great tool we now have to use. Let’s take it one step further. Let’s declare a variable with these values ahead of time:

gross_profit_accounts = ('0170100', '0170200', '0170300')

And then we’ll check if the account is one of these:

if account in gross_profit_accounts:
print "Matched account against account list!"

Again, this code works the exact same way, but we’ve introduced a variable name. Think of the variable as a bit of built-in documentation. We easily could have done this too:

# This is for example purposes, the previous example is the best one (in my opinion)
# Check if the account is one of the gross profit accounts
if account in ('0170100', '0170200', '0170300'):
    print "Matched account using tuple!"

But now without even using a comment, the variable name serves as an expressive explanation of what the code is doing. Again, code is read more than it is written, so I think this is a better solution than using the variable name “gp_accounts” or “gpa” or something that requires a decoder ring.

Think about the English translation of this code now: “If the account is a gross profit account, do this thing.” Also, what if we wanted to test the opposite case (not being a gross profit account)? We could do this:

if account not in gross_profit_accounts:
    print "Account not a gross profit account!"

In English: “If the account is not a gross profit account, do this thing.” It’s like the code documents itself – and that is the best kind of code. Think of code as prose!