Character Encoding Handled the right way

[thorsten_gawantka]

In our daily business we have to handle different types of data-sources and data-sinks. Most of these data-source and data-sinks are UTF-8 aware already. But what if one of these is does not provide data in UTF-8 or does not understand it at all?

 

The PI/PO is only handling UTF-8-Data so how is it possible within the PI to convert the Encoding of a File/Message?

 

In one of our current scenarios we have to handle fixed-length files, which are read from a SFTP server.

(But this guide should also be valid for other scenarios)

 

The configuration is simple, but not obvious since the behaviour of the MessageTransformBean is a little bit strange :wink:

 

Here is what you have to do in the Module-Chain of your channel if you want to convert a message from a supported encoding into UTF-8.

 

Processing Sequence

Number	Module Name	Type	Module Key
1	AF_Modules/MessageTransformBean	Local Enterprise Bean	Codepage
2	AF_Modules/TextCodepageConversionBean	Local Enterprise Bean	Convert
3	AF_Modules/MessageTransformBean	Local Enterprise Bean	Transform
4	localejbs/CallSapAdapter	Local Enterprise Bean	entry

 

Module Configuration

Module Key	Parameter Name	Parameter Value
Codepage	Transform.ContentType	text/plain;charset=ISO-8859-1
Convert	Conversion.charset	utf-8
Transform	Transform.Class	com.sap.aii.messaging.adapter.Conversion
Transform	xml.conversionType	SimplePlain2XML
Transform	xml.documentName	Foo
Transform	xml.documentNamespace	http://example.com
Transform	xml.fieldFixedLength	1,2,3
Transform	xml.fieldNames	ID, Bar, Baz
Transform	xml.processFieldNames	fromConfiguration
Transform	xml.structureTitle	Bar

 

The key in this configuration is to use two different instance of the MessageTransformBean. The first one only sets the codepage, and mime-type, which is then converted by the TextCodepageConversionBean. The sencond one parses the (now utf-8) text into xml.

 

These two steps are necessary since the class "com.sap.aii.messaging.adapter.Conversion" expects its data as utf-8 which is not right if the incomming codepage is not utf-8 but an other one.