Question

In the process of moving some applications from PHP and Delphi to Java-RS some column entries are spoiling the show at this time.

After reading the data from mySQL with JPA into Java Pojos converting the result to XML using JaxB and trying to read back the result to JQuery / jqGrid a failure happens.

In the browser the problem simply shows at “not well formed”.
Looking at the details with the Eclipse XML editor gives the error message:

An invalid XML character (Unicode 0x0) was found in the element content of the document

Now I’d like to proceeed and fix the original data.

1. How would an SQL query look like that looks for the rows that have
   invalid entries?
2. How would an SQL query look like that fixes these
   rows?

Let’s assume the column with the problem is “name” in Table “Customer”

For question #1 I found:
SELECT name from customer where hex(name) like “%00”;

to work. For question #2 I am assuming that update with a left substring might work. I am not sure about the length in this case. It looks like length(name) will return the length including the terminating zero character. Will the update with left(length(name)-1) work correctly in this case?

I am not sure whether backup and restore of the database would keep the current somewhat corrupted database in shape. So it would also be helpful to know how the situation can be reproduced with an insert statement that creates null terminated strings on purpose.

Answer 1

I think you should be able to transform the HEX() with something like

UPDATE customer SET name=UNHEX(REPLACE(HEX(name), '00', ''));

or simply

UPDATE customer SET name=REPLACE(name, CHAR(0), '');