How to fix the xml?

Question

We have the dilema of geting from an outside application to us a file with xml.
We push the files in WebDAV and we have the information in the repository.
Our problem is that the files are like:
< Document >
< Field1> val1 </Field1>
< Field2> <?xml ,,,,,,
&lt;Field3&gt;Value4&lt;/Field3&gt; .....
>
</Field2>
</Field1>
</Document>
Our objective is :
use the WebDAV and read the xml content from the resource_view after been inserted using webdav and copy/insert in a table in our aplication the content of the xml file.
Our Problems:
1)-We have in a field in the document XML where the begin and ending tags are like "&lt;" and "&gt;" also we have the "?xml" on it.
2)- The files are bigger than 4000 chars.
3)- When we try to use any of the xml function it fail.
a)
I have try to read the xml from the repository using the getCLobVal() and convert the information ot a XMLTYPE field in a table. But doesn't work.
We use also replace() on the clob field to fix the "&lt;" , "&gt;" and the "?xml" part but we also have problems.
b)
I have been able to read the files from a directory with bfile.. as clob and fix the xml with the replace and able to write to the tables and do the interpretation of the file in our tables. But how I will be able to write this file to the directory after I have the information in the repository. We don't need an automatic process.
The web application is presenting "path_view" information in a web page where the user pick/click the file to process.
Thank you for your help in advance.
I can submit a real file if email is provided.
Jose.

Answer 1

Your file is not valid XML. The only way that <?xml> can appear inside <Field2> is if it is enclosed in a CDATA Section. If that is the case you should be able to extract() the contents of field2/text(), and then use dbms_xmlgen to convert the contents from escaped XML to real XML...
SQL> var xmltext1 clob
SQL> var xmltext2 clob
SQL> var xmltext3 clob
SQL> --
SQL> set define off
SQL> begin
  2    :xmltext1 := '<T><FIELD1><?xml version="1.0" encoding="UTF-8"?><Test/></FIELD1></T>';
  3    :xmltext2 := '<T><FIELD1><![CDATA[<?xml version="1.0" encoding="UTF-8"?><Test/>]]></FIELD1></T>';
  4    :xmltext3 := '<T><FIELD1><?xml version="1.0" encoding="UTF-8"?><Test/></FIELD1></T>';
  5  end;
  6  /
PL/SQL procedure successfully completed.
SQL> set long 10000
SQL> --
SQL> select xmltype(:xmltext1)
  2    from dual
  3  /
ERROR:
ORA-31011: XML parsing failed
ORA-19202: Error occurred in XML processing
LPX-00209: PI names starting with XML are reserved
Error at line 1
ORA-06512: at "SYS.XMLTYPE", line 272
ORA-06512: at line 1
no rows selected
SQL> select xmltype(:xmltext2)
  2    from dual
  3  /
XMLTYPE(:XMLTEXT2)
<T>
  <FIELD1><![CDATA[<?xml version="1.0" encoding="UTF-8"?><Test/>]]></FIELD1>
</T>
SQL> select xmltype(:xmltext3)
  2    from dual
  3  /
XMLTYPE(:XMLTEXT3)
<T>
  <FIELD1><?xml version="1.0" encoding="UTF-8"?><Te
st/></FIELD1>
</T>
SQL> select extract(xmltype(:xmltext2),'/T/FIELD1/text()').getClobVal()
  2    from dual
  3  /
EXTRACT(XMLTYPE(:XMLTEXT2),'/T/FIELD1/TEXT()').GETCLOBVAL()
<![CDATA[<?xml version="1.0" encoding="UTF-8"?><Test/>]]>
SQL> select extract(xmltype(:xmltext2),'/T/FIELD1/text()').getClobVal()
  2    from dual
  3  /
EXTRACT(XMLTYPE(:XMLTEXT2),'/T/FIELD1/TEXT()').GETCLOBVAL()
<![CDATA[<?xml version="1.0" encoding="UTF-8"?><Test/>]]>
SQL> select xmltype(substr(CDATA,10,length(CDATA)-12))
  2   from (
  3           select extract(xmltype(:xmltext2),'/T/FIELD1/text()').getClobVal() CDATA
  4           from dual
  5        )
  6  /
XMLTYPE(SUBSTR(CDATA,10,LENGTH(CDATA)-12))
<?xml version="1.0" encoding="WINDOWS-1252"?>
<Test/>
SQL> select xmltype(dbms_xmlgen.convert(extract(xmltype(:xmltext3),'/T/FIELD1/text()').getClobVal(),1))
  2    from dual
  3  /
XMLTYPE(DBMS_XMLGEN.CONVERT(EXTRACT(XMLTYPE(:XMLTEXT3),'/T/FIELD1/TEXT()').GETCL
<?xml version="1.0" encoding="WINDOWS-1252"?>
<Test/>
SQL>Note in the above example the assignment to xmltext3 should read
:xmltext3 := '<T><FIELD1> & lt;?xml version="1.0" encoding="UTF-8"? & gt; & lt;Test/ & gt;</FIELD1></T>'; but without the ' ' before and after each '&' character
Message was edited by:
mdrake