When users attempt to parse a 'UTF-8'-encoded BPMN file (while the locale charset is not set as 'UTF-8') using following codes:

    String fileContent = IOUtils.readStringAndClose(new InputStreamReader(new FileInputStream(modelFile), "utf-8"), (int) modelFile.length());
    BpmnModel bpmnModel = new BpmnXMLConverter().convertToBpmnModel(new StringStreamSource(fileContent), false, false); 

the result bpmnModel is wrong because properties of all model elements are parsed in locale charset, Chinese 'gb2312' for example.

I think this error is caused by that StringStreamSource.getInputStream() returns a byte array stream using locale charset encoding, and that is why I add a new two-arguments constructor for StringStreamSource class.

Now, codes could be written as below:

    BpmnModel bpmnModel = new BpmnXMLConverter().convertToBpmnModel(new StringStreamSource(fileContent, "utf-8"), false, false);