"Content is not allowed in prolog" when parsing perfectly valid XML on GAE

I've been beating my head against this absolutely infuriating bug for the last 48 hours, so I thought I'd finally throw in the towel and try asking here before I throw my laptop out the window.

I'm trying to parse the response XML from a call I made to AWS SimpleDB. The response is coming back on the wire just fine; for example, it may look like:

<?xml version="1.0" encoding="utf-8"?> 
<ListDomainsResponse xmlns="http://sdb.amazonaws.com/doc/2009-04-15/">
    <ListDomainsResult>
        <DomainName>Audio</DomainName>
        <DomainName>Course</DomainName>
        <DomainName>DocumentContents</DomainName>
        <DomainName>LectureSet</DomainName>
        <DomainName>MetaData</DomainName>
        <DomainName>Professors</DomainName>
        <DomainName>Tag</DomainName>
    </ListDomainsResult>
    <ResponseMetadata>
        <RequestId>42330b4a-e134-6aec-e62a-5869ac2b4575</RequestId>
        <BoxUsage>0.0000071759</BoxUsage>
    </ResponseMetadata>
</ListDomainsResponse>

I pass in this XML to a parser with

XMLEventReader eventReader = xmlInputFactory.createXMLEventReader(response.getContent());

and call eventReader.nextEvent(); a bunch of times to get the data I want.

Here's the bizarre part -- it works great inside the local server. The response comes in, I parse it, everyone's happy. The problem is that when I deploy the code to Google App Engine, the outgoing request still works, and the response XML seems 100% identical and correct to me, but the response fails to parse with the following exception:

com.amazonaws.http.HttpClient handleResponse: Unable to unmarshall response (ParseError at [row,col]:[1,1]
Message: Content is not allowed in prolog.): <?xml version="1.0" encoding="utf-8"?> 
<ListDomainsResponse xmlns="http://sdb.amazonaws.com/doc/2009-04-15/"><ListDomainsResult><DomainName>Audio</DomainName><DomainName>Course</DomainName><DomainName>DocumentContents</DomainName><DomainName>LectureSet</DomainName><DomainName>MetaData</DomainName><DomainName>Professors</DomainName><DomainName>Tag</DomainName></ListDomainsResult><ResponseMetadata><RequestId>42330b4a-e134-6aec-e62a-5869ac2b4575</RequestId><BoxUsage>0.0000071759</BoxUsage></ResponseMetadata></ListDomainsResponse>
javax.xml.stream.XMLStreamException: ParseError at [row,col]:[1,1]
Message: Content is not allowed in prolog.
    at com.sun.org.apache.xerces.internal.impl.XMLStreamReaderImpl.next(Unknown Source)
    at com.sun.xml.internal.stream.XMLEventReaderImpl.nextEvent(Unknown Source)
    at com.amazonaws.transform.StaxUnmarshallerContext.nextEvent(StaxUnmarshallerContext.java:153)
    ... (rest of lines omitted)

I have double, triple, quadruple checked this XML for 'invisible characters' or non-UTF8 encoded characters, etc. I looked at it byte-by-byte in an array for byte-order-marks or something of that nature. Nothing; it passes every validation test I could throw at it. Even stranger, it happens if I use a Saxon-based parser as well -- but ONLY on GAE, it always works fine in my local environment.

It makes it very hard to trace the code for problems when I can only run the debugger on an environment that works perfectly (I haven't found any good way to remotely debug on GAE). Nevertheless, using the primitive means I have, I've tried a million approaches including:

  • XML with and without the prolog
  • With and without newlines
  • With and without the "encoding=" attribute in the prolog
  • Both newline styles
  • With and without the chunking information present in the HTTP stream

And I've tried most of these in multiple combinations where it made sense they would interact -- nothing! I'm at my wit's end. Has anyone seen an issue like this before that can hopefully shed some light on it?

Thanks!


Solution 1:

The encoding in your XML and XSD (or DTD) are different.
XML file header: <?xml version='1.0' encoding='utf-8'?>
XSD file header: <?xml version='1.0' encoding='utf-16'?>

Another possible scenario that causes this is when anything comes before the XML document type declaration. i.e you might have something like this in the buffer:

helloworld<?xml version="1.0" encoding="utf-8"?>  

or even a space or special character.

There are some special characters called byte order markers that could be in the buffer. Before passing the buffer to the Parser do this...

String xml = "<?xml ...";
xml = xml.trim().replaceFirst("^([\\W]+)<","<");

Solution 2:

I had issue while inspecting the xml file in notepad++ and saving the file, though I had the top utf-8 xml tag as <?xml version="1.0" encoding="utf-8"?>

Got fixed by saving the file in notpad++ with Encoding(Tab) > Encode in UTF-8:selected (was Encode in UTF-8-BOM)