Null Behavior: Dealing with missing Attributes/values
To deal with missing values we can either use the built-in Null Behavior feature of Attribute Maps, or we can detect and handle this yourself.
Let's start by configuring Null Behavior. Do this by right-clicking
on the 'Read_CSV_File' Connector in the Attribute Maps panel (not
in the AL components tree-view) and selecting Null Behavior from
the context menu1.
This brings up the Null Behavior dialog where we can configure
both how null is defined – which can vary depending on
the type of source you are reading from – as well as how this situation
should be handled. By default, null means that an Attribute
is missing, and the default handling is to remove this Attribute from
the mapping operation. The end result of this is that no Attribute
with the specified name will be found in the receiving Entry.
Once in the Null Behavior dialog, use the radio buttons on the
right to define null as an empty string value, and then
those on the left to specify a default value of "* missing *" to be
returned in this case.
Re-run the AssemblyLine and refresh the browser window displaying
the contents of Output.xml. The entry for 'Roger'
should now have the special null value ("* missing *")
for 'Title' and 'LastName'.
This is somewhat better – at least it's a conscious choice. However,
sometimes an entry is just too incomplete to continue processing.
In our scenario you need at least values for 'FirstName' and 'LastName'
in order to compute 'FullName', so you will now add filtering logic
to the AssemblyLine to ensure that all entries fulfill this requirement.
Start by clicking the Insert Component button
again and then choosing Control/Flow Components from
the radio buttons at the left. Then select 'IF' in the list, name
it 'Incomplete data'2 and press Finish.
Now drag this IF branch above your output Connector. The IF branch
editor area lets you add your conditions.
Here you have the option of adding simple conditions or writing
a snippet of Script – or both. Note the Match All checkbox
that decides whether your Conditions (simple and scripted) are evaluated
with an implied OR between them when this is unchecked,
or with AND when it is selected.
Add a simple condition by pressing the Add button.
Then pick the 'First' Attribute from the leftmost drop-down and then
the 'has value(s)' operator. Negate this condition by toggling the not column
value. Nothing needs to be specified in the right-most field when
using the 'has value(s)' operator. Now add a similar has no
values condition for the Attribute named 'Last' as well. Finally,
make sure the Match All checkbox is unchecked
– which implies Match Any – so that the lack of values for
either Attribute will trigger this branch.
An IF Branch diverts the execution flow of the AssemblyLine whenever
the specified Conditions evaluate to true. In your case,
processing will continue to those components placed under the branch
if either the 'First' or 'Last' Attributes do not have values assigned
to them – or even if either does not exist in the Work Entry. In other
words, the 'has value(s)' operator also includes the check for 'exists'.
Now expand the IF branch and double-click on the placeholder displayed
under it to insert a component here. In the Choose Component dialog,
click on the radio button for Scripts, select
the one called 'Script' and press Finish. Rename
this Script component (also called an 'SC') to 'Write to log' and
then enter the following snippet of JavaScript in it3:
The task variable used here references the AssemblyLine
itself, and it provides you with a number of useful functions like
the logmsg() method. This scripted call causes the
specified text message to be written to your log output.
To make log output even more informative, let's include the
current contents of the Work Entry as well. Do this by right-clicking
on the IF branch, selecting Add Component... and
again choosing the radio button for Scripts.
This time select the pre-defined Script component labeled 'Dump Work
Entry'.
Finally, you must instruct the AL to stop the current cycle at
this point and return control to the Iterator Connector so that it
can read in the next CSV line, effectively filtering the current Entry
from the XML output. Unless you specify this behavior yourself then
control will continue to the first component after the IF branch.
So once again you must right-click on your IF branch and then on Add
Component... Choose Scripts and
then select the SC called 'Exit Flow'. Now your AL should look like
this:
Before running the AssemblyLine again you will need to open the
Null Behavior dialog once more and restore both the default definition
and behavior selections – otherwise your IF Branch conditions will
never evaluate to true.
Now when you run the AssemblyLine again you will see your message
followed by the Work Entry dump:
From the statistics we can see that your IF 'Data incomplete'
branch was true once, resulting in only six nodes being added to your
XML output. If you again refresh the Data Browser window then you
will see that 'Roger' is indeed gone.
1 Or we can select the Connector itself in the
AL component list; press the More... button
at the top of the Input Map and choose 'Null Behavior' there.2 You were previously encouraged to name Connectors
as you would a script variable. This also applies to Function components.
However, it is less important for Attribute Map components, Branches,
Loops and Scripts, since these are seldom accessed directly from script.
In this guide higher priority has been given to making the AssemblyLine
easy to read.3 The Script editor
provides a feature called code-completion that shows
which options you have. For example, type "tas" in the Script editor
and then press the Ctrl + Space key combination to open the code-completion
drop-down, which should provide a single option: task.
If you press Enter then this choice is selected and entered in your
script. Now type the period key (.) so your script becomes "task."
and wait just a moment; You will see a new code-completion drop-down
appear, this time with a list of all the methods and properties that we can access in the task object.
task.logmsg("*** Skipping incomplete entry");
Parent topic:
Introducing SDI