[OPENJPA-2552] When reading orm.xml encoding defined in prolog should be taken into account

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.3.0
• Assignee: Unassigned
• Reporter: Ognjen Blagojević
• Created: 24/Dec/14 13:58
• Updated: 24/Dec/14 13:58

Description

File orm.xml may contain JPQL or SQL queries with string literals with non-ASCII characters, e.g.

<named-native-query name="test">
<query>
select 'ћирилица'
</query>
</named-native-query>

Note the string literal with cyrillic letters.

If JVM variable file.encoding is set to something different (e.g. -Dfile.encoding=ISO-8859-1) from encoding of the file, specified in prolog of orm.xml (e.g. UTF-8), OpenJPA will respect file.encoding setting, which will demage string literals.

According to XML specification, OpenJPA should respect encoding as specified in prolog.

The problematic method seems to be: org.apache.openjpa.lib.meta.XMLMetaDataParser. parse(MetaDataIterator itr, boolean stopFirst), which opens InputStreamReader with default encoding from file.encoding variable.