FileAdapter write mode - Dynamic filename, write down simple text structure

Question

Hi,
I tried to use the FileAdapter in a BPEL Process to write down and existing structure as String, comming from Database (CLOB), which has a defined structure.
My first problem is, that I assign an output filename through the output message header variable, but it does not work. Instead always the name from the Filename convention is taken (wizard like po_%SEQ%.txt).
My second problem is, that the FileAdapater destroyes the format during the write procedure. E.g.: The String structure is as follows:
MKKOPF EF585773 07.05.2009 1 XXX Spielapparate u. Restaurant- betriebsgmbH 40031356 Brehmstraße 21 1110 Wien 900000585773 EUR ML
MKMAIL [email protected]
MKEINZ EF4428643 28.06.2009 Test Card Casino XXX Gesamtausgabe Textteil Allgemein Großanzeige 4C 1 ST 1 489,00 489,00 GES
MKPOS. EF4428644 26.07.2009PRVB 948,66 / 10 / +5% WA
a.s.o, . So you can see, its quite a simple format, which has to stay like that!
First I tried to use the opaque mode for the fileoutput format, but I was stopped, because of the '/' characters in the source data. So I used the native builder to create a xml schema (I used deliminated file, multiple records, deliminated by white spaces, tabs and spaces). To identify the multiple records, I have chosen the end of line (eol). Except EOF and EOL there is no other choice, like CRLF,\r,\n!
However, testing my native format, result in writting a file with containing a single line with the original data. So the original line breaks are gone, and therefore the system who should process it further gets an parsing error!
The simple question is, how can I simple write down string data (containing German special chars, like Öäü --> ISO8859-1 and also XML reserved values like /) through the file adapter without destroying its structure?
Second, how can I pass dynamically the filename, as using the output header values does not work?
I hope some specialists are out there.
Thanks & BR,
Peter

Answer 1

Hi James,
here what I am doing.
I read some records from the database throught the database adapter. The returned records representing strings, which should be written, line by line separated by 0xA (\n) to a single file.
Also for the file write we use a Oracle SOA Adapter --> the FileAdapter.
We discussed quiet some time about opaque schema and native builder. So on my understanding only the native builder version works. Otherwise I get an error, because of the contained slashes in the string record.
What I am searching for is an easy way for aggregating this database row strings and append it with 0xA (\n).
With XPath functions, like create-deliminated-string(node, '#xA') I had no success. Adhering to XML the signs #xA should represent 0xA, but it does not work.
Also in using the encodeLineTerminators 
 is not working.
So I had really write a long bpel construct, combined with a Java Code Emeeding to do such a simple thing:
Here is the interessting part (hiden the while loop, index handling and final fileadapter write call details):
<while name="While_1"
condition="($VariableSapBillingJobs > 0) and ($VariableIndex &lt;= $VariableSapBillingJobs)">
<sequence name="Sequence_4">
<sequence name="Sequence_4">
<assign name="Assign">
<copy>
<from expression="concat(bpws:getVariableData('VariableBuffer'), bpws:getVariableData('InvokeSapDataSelection_eBillingSapDB4Select_OutputVariable','eBillingSapDB4SelectOutputCollection','/ns8:eBillingSapDB4SelectOutputCollection/ns8:eBillingSapDB4SelectOutput[$VariableIndex]/ns8:LINE'))"/>
<to variable="VariableBuffer"/>
</copy>
</assign>
*<bpelx:exec name="JavaAppendCRLF" language="java"*
version="1.5">
<![CDATA[String buffer = (String)getVariableData("VariableBuffer");    
buffer += "\n";     
setVariableData("VariableBuffer", buffer);]]>
</bpelx:exec>
<assign name="AggregateStructure">
<copy>
<from expression="$VariableIndex + 1"/>
<to variable="VariableIndex"/>
</copy>
</assign>
</sequence>
</sequence>
</while>
There must be a simpler way like that!
Do you or somebody else a simpler way? Doing it with the FileAdapter native builder, or at least with a XPath function?
Furthermore my next homework is, the inverse operation. I will read a file, this time terminated with 0xD 0xA, and must translate it to XML to use it further in my bpel process.
Also here I have not the idea, where the fileadapter supports me.
Thanks & BR,
Peter