Special Characters in xUnit Report

[tmbrbr]

When running the JavaScript test suite in the browser context jstestbrowser, it is possible to specify a flag to produce xUnit XML output, i.e.

./mach jstestbrowser --log-xunit=jsbrowsertest_output.xml --headless

However, some of the tests contain ASCII control characters in their names (generated automatically from the test content), for example, small-codepoints.js

var gTestfile = 'small-codepoints.js';
//-----------------------------------------------------------------------------
var BUGNUMBER = 554079;
var summary = 'JSON.parse should reject U+0000 through U+001F';

print(BUGNUMBER + ": " + summary);

/**************
 * BEGIN TEST *
 **************/

for (var i = 0; i <= 0x1F; i++)
  testJSON('["a' + String.fromCharCode(i) + 'c"]', true);

The code responsible for creating the xml is xunit.py which creates an XML tree, containing the control characters. Minidom is then used to parse the output, but throws a parser error when it finds the control chars.

I pushed a workaround (410b044) which removed control chars before passing to minidom, which seems to work.

This issue is open as a reminder to raise a bug report with firefox as it will also be present there.