Multiple Feature Attribute Support of the AttributeCreator in FME 2013 SP2

The Multiple Feature Attribute Support functionality has been added to the AttributeCreator in FME 2013 SP2, it realizes accessing freely to attributes of backward / forward features in a sequence of features. Using this functionality, for example, the Fibonacci sequence can be created easily by a Counter and an AttributeCreator with the settings like:

Count Output Attribute: _fibonacci
Count Start: 0

<Mutliple Feature Attribute Support>
Number of Prior Features: 2
<Attribute To Set>
Attribute Name: _fibonacci
Value (Conditional):
  Test Condition: 1 < @Value(feature[0]._fibonacci)
  Output Value: @Evaluate(@Value(feature[-2]._fibonacci)+@Value(feature[-1]._fibonacci))

After this, using an AttributeAccumulator and a ListConcatenator, got a correct Fibonacci sequence: 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, ...

The point is that we can access any attributes of prior / subsequent  features with the syntax of feature[n].<attribute name>n is an integer value which indicates the relative position of a feature based on the position of the current feature in the sequence. Negative n means a prior feature, positive n means a subsequent feature, and naturally 0 means the current feature. The range of available n value will be determined according to "Number of Prior Features" and "Number of Subsequent Features" parameters.

Does the Fibonacci sequence seem useless?
OK, see:
Great! Multiple Feature Attribute Support in AttributeCreator FME 2013 SP2

In FME 2013 SP3, we don't need to use the syntax of feature[0].<attribute name> for the current feature attribute. Just <attribute name> indicates the current feature attribute in condition expressions. For example:
   Test Condition: 1 < @Value(_fibonacci)

Next: Use Case of Multiple Feature Attribute Support


BulkAttributeRenamer + AttributeAccumulator creates single record from multiple records which are storing attributes of a feature dispersedly

Sometimes, we have a table that stores attributes of a feature in multiple rows dispersedly like this:
id, attrName, attrValue
001, aaa, 1
001, bbb, 2
002, aaa, 3
002, bbb, 4

How to transform this table to the following structure?
id, aaa, bbb
001, 1, 2
002, 3, 4

One possible way is to accumulate attributes grouped by id, and then to restructure the table using a Python script:
import fmeobjects
def restructureTable(feature):
    names = feature.getAttribute('_list{}.attrName')
    values = feature.getAttribute('_list{}.attrValue')
    for name, value in zip(names, values):
        feature.setAttribute(name, value)

It works well but is a little inefficient, because it needs a list attribute which will not be used afterward. But I didn't know better ways.
Now,  I've learned a solution using the BulkAttributeRenamer with the following parameter settings:
Selected Attributes: attrValue
Action: Regular Expression Replace
Text to Find: .*
String: attrName

After renaming, I can create a table having the preferable schema by the AttributeAccumulator without creating list attributes.

Community > RecordBuilder / CSV import
Community > Attribute renaming using attribute values


How to specify multiple xfMaps to XML reader

To specify multiple xfMaps to the XML reader (FME 2013), the paths should be delimited with semicolons like:
"C:\tmp\xfmaps\sample01_2.xml ; C:\tmp\xfmaps\sample01_1.xml"
Nobody noticed this until now?
Community > How to specify multiple xfMap files to an XML reader

2014-09-29: As I mentioned in the post linked above, this issue has been resolved in FME 2014. In FME 2014, you don't need delimiters when specifying multiple xfMap files to an XML reader.