[AMQ-8398] 4-byte Unicode message from JMS to STOMP will be corrupted

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 5.16.3
• Component/s: Broker, STOMP, Transport
• Assignee: Jean-Baptiste Onofré
• Reporter: Simon Lundstrom
• Created: 06/Oct/21 15:05
• Updated: 28/Jun/23 15:38

Description

When sending a message from:
JMS producer to STOMP consumer
or
STOMP producer to JMS consumer
which contains a 4-byte unicode code points e.g. https://unicode-table.com/en/1F5A4/ there is a corruption of the message.
In the JMS to STOMP case the code point gets converted to:
ef bf bd ef bf bd when it should be f0 9f 96 a4.
and in the STOMP to JMS case the JMS client throws an exception:

Exception in thread "main" javax.jms.JMSException: java.io.UTFDataFormatException
        at org.apache.activemq.util.JMSExceptionSupport.create(JMSExceptionSupport.java:72)
        at org.apache.activemq.command.ActiveMQTextMessage.decodeContent(ActiveMQTextMessage.java:104)
        at org.apache.activemq.command.ActiveMQTextMessage.getText(ActiveMQTextMessage.java:84)
        at testkonsument.App.JMS(App.java:86)
        at testkonsument.App.main(App.java:42)
Caused by: java.io.UTFDataFormatException
        at org.apache.activemq.util.MarshallingSupport.convertUTF8WithBuf(MarshallingSupport.java:389)
        at org.apache.activemq.util.MarshallingSupport.readUTF8(MarshallingSupport.java:358)
        at org.apache.activemq.command.ActiveMQTextMessage.decodeContent(ActiveMQTextMessage.java:101)
        ... 3 more

Using 4-byte unicode points
from STOMP to STOMP
or
from JMS to JMS
is not a problem, both works and does not corrupt the code point.

Note that 2- (e.g. https://unicode-table.com/en/00F6/) or 3-byte (e.g. https://unicode-table.com/en/2614/) Unicode code points does NOT get corrupted, even if the same message includes a 4-byte Unicode code point.