FW: Content is not allowed in prolog

["Mitchell, Steven C" <St...@umb.com>]

The cause of the "Content is not allowed in prolog" error we had on AIX
turned out to be a bad application properties file on that box.  The
property specifying the location of the sql mapping config file got run
together with a comment line above it.  My theory is that when the
property value was unsuccessfully substituted in the dao.xml file, the
bad sql mapping config file location was given to the parser which
resulted in the misleading "Content is not allowed in prolog" error.  It
was not that it couldn't parse the sql config mapping file, it actually
couldn't find it.  I don't know why we didn't see a different error
message.

We opened a ticket with IBM while we tried to figure this out.  IBM had
the following say about the iBatis code.  Although IBM was mistaken
about the cause of the error, they pointed out some things about the
iBatis code that might be worth looking into:

"I've taken a deeper look and it appears that iBatis is at fault. 

In the PMR log I noticed the following trace: 

[6/21/05 9:33:41:994 CDT] 53749a3f WebGroup      E SRVE0026E: [Servlet 
   Error]-[Error while configuring DaoManager.  Cause: 
   com.ibatis.common.exception.NestedRuntimeException: Error occurred. 
   Cause: com.ibatis.common.xml.NodeletException: Error parsing XML. 
   Cause: org.xml.sax.SAXParseException: Content is not allowed in
prolog. 
   Caused by: org.xml.sax.SAXParseException: Content is not allowed in 
   prolog. 
   Caused by: com.ibatis.common.xml.NodeletException: Error parsing XML.

   Cause: org.xml.sax.SAXParseException: Content is not allowed in
prolog. 
   . 
   6/22/05 13:27:46:202 CDT] 55951fa2 RequestProces W 
   org.apache.struts.action.RequestProcessor  Unhandled Exception
thrown: 
   class com.ibatis.dao.client.DaoException 

and then found the source (in Apache) in the
com.ibatis.common.xml.NodeletParser [1] class which throws the
com.ibatis.common.xml.NodeletException [2].  NodeletParser, specifically
the method NodeletParser.createDocument(Reader) passes a java.io.Reader
to the parser as an input.  This is the problem.  When the parser reads
from an InputStream or from a URL it determines the encoding of the
document from the XML declaration (i.e. <?xml version="1.0"
encoding="ISO-8859-1"?>) and then chooses the appropriate java.io.Reader
to use for that encoding.  iBatis doesn't give the parser the chance to
pick the proper java.io.Reader for the document since it provides the
Reader directly. 

The java.io.Reader which is passed to the parser is a
java.io.InputStreamReader created by
com.ibatis.common.resources.Resources [3] from
SqlMapConfigParser.addSqlMapNodelets() [4].  The javadoc for the
constructor of InputStreamReader [5] which the Resources class uses
says: "Create an InputStreamReader that uses the default charset."  I
guess this is working on Windows because the default encoding just
happens to be the same as (or is compatible with) the encoding of the
document.  On AIX it would seem like this is not the case, so it fails.
Because the default encoding varies from platform to platform creating
an InputStreamReader with the default encoding is always a bug waiting
to happen. 

[1]
http://svn.apache.org/viewcvs.cgi/ibatis/trunk/java/mapper/mapper2/src/c
om/ibatis/common/xml/NodeletParser.java?rev=180318&view=markup 
[2]
http://svn.apache.org/viewcvs.cgi/ibatis/trunk/java/mapper/mapper2/src/c
om/ibatis/common/xml/NodeletException.java?rev=180318&view=markup 
[3]
http://svn.apache.org/viewcvs.cgi/ibatis/trunk/java/mapper/mapper2/src/c
om/ibatis/common/resources/Resources.java?rev=180318&view=markup 
[4]
http://svn.apache.org/viewcvs.cgi/ibatis/trunk/java/mapper/mapper2/src/c
om/ibatis/sqlmap/engine/builder/xml/SqlMapConfigParser.java?rev=180318&v
iew=markup 
[5]
http://java.sun.com/j2se/1.4.2/docs/api/java/io/InputStreamReader.html#I
nputStreamReader(java.io.InputStream) 



Thanks,
Paul McAdams 
IBM WebSphere Application Server Support"

-----Original Message-----
From: Larry Meadors [mailto:larry.meadors@gmail.com] 
Sent: Wednesday, June 22, 2005 11:03 AM
To: user-java@ibatis.apache.org
Subject: Re: Content is not allowed in prolog


This really is not an iBATIS issue, but I suspect you must need to
somehow edit the XML in a plain text editor, and make sure there are no
unicode charaters in it, and that the very first line begins with <?xml
- on the first line with no spaces before it.

On 6/22/05, Mitchell, Steven C <St...@umb.com> wrote:
>  
> We recently starting developing with iBatis (I have used it before at 
> other sites).  It runs fine on a WebSphere 5.1 under Rational 
> Application Developer 6.0 on Windows XP; however, when we try to 
> deploy to WebSphere 5.1 on AIX iBatis will not load and we get the 
> following error:
>  
> 
> Error 500: Error while configuring DaoManager. Cause:
> com.iBatis.common.exception.NestedRuntimeException: Error occurred. 
> Cause: com.ibatis.common.xml.NodeletException:
> Error parsing XML. Cause: org.xml.sax.SAXParseException: Content is 
> not allowed in prolog. Caused by: org.xml.sax.SAXParseException: 
> Content is not allowed in prolog. Caused by:
> com.ibatis.common.xml.NodeletException: Error parsing XML.
> Cause: org.xml.sax.SAXParseException: Content is not allowed in 
> prolog. Caused by: org.xml.sax.SAXParseException: Content is not 
> allowed in prolog. One of my team members found this link about a bug 
> in RAD 6.0 where it writes a bad Byte Offset Mark (BOM) in XML files 
> (http://forum.java.sun.com/thread.jspa?threadID=567285&tstart=90),
> so we implemented the recommended circumvention on all the iBatis XML 
> files and switched from UTF-8 to ISO-8859-1 encoding. We still get the

> error.
>   
> Has anybody else had this problem?  Any suggestions?