Content is not allowed in prolog.

[lewis.watson]

Hello,

I'm in the process of setting up a REST web API as a data source in JBoss Data Virtualization. I created a API using Python REST API Framework — Eve 0.4-dev documentation but ran into difficulties when previewing my view.

SELECT
        A.location AS location, A.role AS role, A.firstname AS firstname, A.born AS born, A.lastname AS lastname
    FROM
        (EXEC Eve_People_source.invokeHttp('GET', null, 'http://eve-demo.herokuapp.com/people', 'TRUE')) AS f, XMLTABLE('/resource/resource' PASSING XMLPARSE(DOCUMENT f.result) COLUMNS location string PATH 'location/text()', role string PATH 'role/text()', firstname string PATH 'firstname/text()', born string PATH 'born/text()', lastname string PATH 'lastname/text()') AS A

SQL Results

select * from "Eve_People_view"."people"
TEIID30151 Remote org.teiid.core.TeiidProcessingException: TEIID30151 Error building Source for context item.
Elapsed Time:  0 hr, 0 min, 1 sec, 896 ms.

Server Console

15:55:56,751 ERROR [stderr] (Worker1_QueryProcessorQueue3) Error on line 1 column 1 
15:55:56,751 ERROR [stderr] (Worker1_QueryProcessorQueue3)   SXXP0003: Error reported by XML parser: Content is not allowed in prolog.
15:55:56,754 WARN  [org.teiid.PROCESSOR] (Worker0_QueryProcessorQueue2) TEIID30020 Processing exception for request 0NOkK/7jXj13.0 'TEIID30151 Error building Source for context item.'. Originally TeiidProcessingException 'Content is not allowed in prolog.' org.apache.xerces.util.ErrorHandlerWrapper.createSAXParseException(Unknown Source). Enable more detailed logging to see the entire stacktrace.

I believe the problem comes down to a missing line in the XML response from http://eve-demo.herokuapp.com/people.

It currently returns

<resource href="eve-demo.herokuapp.com/people" title="people">
     <link rel="parent" href="eve-demo.herokuapp.com" title="home"/><resource href="eve-demo.herokuapp.com/people/539fdea42b996e00026114fb" title="person">
...

but I think Teiid wants

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<resource href="eve-demo.herokuapp.com/people" title="people">
     <link rel="parent" href="eve-demo.herokuapp.com" title="home"/><resource href="eve-demo.herokuapp.com/people/539fdea42b996e00026114fb" title="person">
...

I don't know how common it is for REST API's to drop the "<?xml..." line but I think ideally Teiid should be able to handle it. Is there something I can do at the Teiid end or will I have to rely on the API maintainer to provide 100% valid XML?

Thanks in advance!

Lewis

[lewis.watson]

My theory about the missing XML deceleration may be wrong...

I set up a REST api at http://95.138.165.64:5002/cases that returns

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<resource href="/cases" title="cases">
    <link rel="parent" href="" title="home"/>
    <_meta>
        <max_results>25</max_results>
        <total>5</total>
        <page>1</page>
    </_meta>
    <resource href="/cases/53a807f61571734c148fc8c9" title="case">
        ...

    </resource>
    ...
</xml>

And with a transformation

SELECT
        A.status AS status, A.received AS received, A.archived AS archived, A.destruction_date AS destruction_date, A.registered AS registered, A.review_requested AS review_requested, A.complexity_rating AS complexity_rating, A.finalised AS finalised, A.crest_number AS crest_number, A.location AS location, A.type AS type
    FROM
        (EXEC RackspaceEveCases_source.invokeHttp('GET', null, 'http://95.138.165.64:5002/cases', 'TRUE')) AS f, XMLTABLE('/resource/resource' PASSING XMLPARSE(DOCUMENT f.result) COLUMNS status string PATH 'status/text()', received string PATH 'received/text()', archived string PATH 'archived/text()', destruction_date string PATH 'destruction_date/text()', registered string PATH 'registered/text()', review_requested string PATH 'review_requested/text()', complexity_rating string PATH 'complexity_rating/text()', finalised string PATH 'finalised/text()', crest_number string PATH 'crest_number/text()', location string PATH 'location/text()', type string PATH 'type/text()') AS A

I still get a "Content is not allowed in prolog" error

12:05:55,408 WARN  [org.teiid.PROCESSOR] (Worker4_QueryProcessorQueue13) TEIID30020 Processing exception for request KiWHPrV70/ln.0 'TEIID30151 Error building Source for context item.'. Originally TeiidProcessingException 'Content is not allowed in prolog.' org.apache.xerces.util.ErrorHandlerWrapper.createSAXParseException(Unknown Source). Enable more detailed logging to see the entire stacktrace.

I've left my example API running at http://95.138.165.64:5002/cases if that helps. It will return XML or JSON depending on the request header.

Regards

Lewis

[shawkins]

Using invokeHttp, I'm seeing the json result back - which will give the prolog error that you are seeing.  Are you needing an accepts or some other header value to be set?

[lewis.watson]

Hi Steven, thanks for looking into this for me.

How are you able to look at the reply? I've been looking around the console output but I can't see anything about the raw response from the server.

Looking at the documentation for invokeHTTP I see that it can take a "contentType" string parameter

Procedure invokeHttp(action in STRING, request in OBJECT, endpoint in STRING, contentType out STRING) returns BLOB

So I tried changing my invokeHttp call to be

EXEC RackspaceEveCases_source.invokeHttp(action => 'GET', endpoint => 'http://95.138.165.64:5002/cases', stream => TRUE, contentType => 'application/xml')

But I still get the prolog error.

08:24:31,003 ERROR [stderr] (Worker3_QueryProcessorQueue7) Error on line 1 column 1 
08:24:31,005 ERROR [stderr] (Worker3_QueryProcessorQueue7)   SXXP0003: Error reported by XML parser: Content is not allowed in prolog.
08:24:31,008 WARN  [org.teiid.PROCESSOR] (Worker2_QueryProcessorQueue6) TEIID30020 Processing exception for request eqvjbAqkH4ob.0 'TEIID30151 Error building Source for context item.'. Originally TeiidProcessingException 'Content is not allowed in prolog.' org.apache.xerces.util.ErrorHandlerWrapper.createSAXParseException(Unknown Source). Enable more detailed logging to see the entire stacktrace.

In my previous attempts to get this working I disabled JSON entirely from the Python-eve REST side. However, without viewing the raw response I can't rule out JSON still sneaking into the mix. For now I've left JSON enabled on http://95.138.165.64:5002/cases as I need Teiid to be able set the contentType.

Regards

Lewis

[shawkins]

> How are you able to look at the reply? I've been looking around the console output but I can't see anything about the raw response from the server.

The response isn't logged, but you can just select the invokeHttp result and inspect it in a client.

> So I tried changing my invokeHttp call to be

The output parameter is for the result value, it is not an input.  I think you'll have to rely on something like:  https://issues.jboss.org/browse/TEIID-3016