copyEventFromReader() ArrayIndexOutOfBoundsException

[dan2097]

The issue can be reproduced by (I attach test.txt):

WstxInputFactory inputFactory = new WstxInputFactory();
WstxOutputFactory outputFactory = new WstxOutputFactory();
String text = FileUtils.readFileToString(new File("test.txt"), "UTF-8");
ByteArrayOutputStream bios = new ByteArrayOutputStream();
try {
  XMLStreamReader2 reader = (XMLStreamReader2) inputFactory.createXMLStreamReader(new StringReader(text));
  XMLStreamWriter2 writer = (XMLStreamWriter2) outputFactory.createXMLStreamWriter(bios, "UTF-8");
  while (reader.hasNext()) {
	reader.next();
	writer.copyEventFromReader(reader, false);
  }
} catch (Exception e) {
  e.printStackTrace();
}

and produces:

java.lang.ArrayIndexOutOfBoundsException: -1
	at com.ctc.wstx.sr.BasicStreamReader.readAndWriteText(BasicStreamReader.java:5185)
	at com.ctc.wstx.sr.BasicStreamReader.getText(BasicStreamReader.java:1431)
	at com.ctc.wstx.sw.BaseStreamWriter.copyEventFromReader(BaseStreamWriter.java:857)
	at test.WoodstoxTest.main(WoodstoxTest.java:27)

To produce the issue all of the following seem to need to be true:

• copyEventFromReader is called with a second parameter of false rather than true
• The root element has a lang attribute
• The element text content is over a certain length

test.txt

EDIT: I reproduced this issue on Woodstox 4.4.1 and 6.0.2

[cowtowncoder]

Thank you for reporting the problem.

[cowtowncoder]

Turns out to be another boundary condition; also required use of \r as linefeed (MacOS I think?) to trigger. Fixed, will be in 6.0.3 once I release that (might be time to do it tonight, 3rd fix).

[dan2097]

Thanks for the quick fix and release. I hadn't actually noticed that \r was being used to indicate new lines!
\r was used by legacy versions of MacOS, but current versions use \n.

[cowtowncoder]

Ah. Makes sense that Macs have switched, perhaps with MacOS X when switching to unix kernel... I just remembered that once upon a time something was using \r. And XML definitely supports 3 linefeed combinations (\n, \r, \r\n).

Also: 6.0.3 should be available now.