[AXIS-2908] Apache Axis fails to handle non Basic Multilingual Plane characters(U+10000 and above) while creating SOAP request

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.4
• Component/s: Serialization/Deserialization
• Labels: charset xml-rpc
• Environment: OS - CentOS; Software Platform - JDK 7
• Assignee: Unassigned
• Reporter: Siddhesh Sundar Toraskar
• Created: 10/Feb/15 08:56
• Updated: 18/Jun/15 21:32

Description

While creating SOAP request, if we have nonBMP characters(e.g. EMOJIs), they(EMOJIs) are not properly inserted in XML.

It seems that my content which is UTF-8 will be encoded in UTF-16 Java String (default) once program receives it.

Apache Axis library that we are using then take those UTF-16 Java Strings and try to convert back into UTF-8 to create a XML before sending. It fails whenever I send a 4-byte Emoji (:grin UTF-8 character. I found that any UTF-8 4-byte character will be represented as surrogate pair in UTF-16. I suspect in that case Axis parser not able to understand surrogate pair and not able to convert into valid UTF-8 encoding.

As result, while UTF-8 is specified, these EMOJIs appear in UTF-16 form which actually corrupts them because they are then incorrectly processed.