diff --git a/common/lib/capa/capa/capa_problem.py b/common/lib/capa/capa/capa_problem.py
index c2bdeadc21..08a223f609 100644
--- a/common/lib/capa/capa/capa_problem.py
+++ b/common/lib/capa/capa/capa_problem.py
@@ -555,6 +555,13 @@ class LoncapaProblem(object):
Used by get_html.
'''
+ if not isinstance(problemtree.tag, basestring):
+ # Comment and ProcessingInstruction nodes are not Elements,
+ # and we're ok leaving those behind.
+ # BTW: etree gives us no good way to distinguish these things
+ # other than to examine .tag to see if it's a string. :(
+ return
+
if (problemtree.tag == 'script' and problemtree.get('type')
and 'javascript' in problemtree.get('type')):
# leave javascript intact.
diff --git a/common/lib/capa/capa/tests/test_html_render.py b/common/lib/capa/capa/tests/test_html_render.py
index 9bc326d7b9..8e343ee1cf 100644
--- a/common/lib/capa/capa/tests/test_html_render.py
+++ b/common/lib/capa/capa/tests/test_html_render.py
@@ -226,6 +226,26 @@ class CapaHtmlRenderTest(unittest.TestCase):
span_element = rendered_html.find('span')
self.assertEqual(span_element.get('attr'), "TEST")
+ def test_xml_comments_and_other_odd_things(self):
+ # Comments and processing instructions should be skipped.
+ xml_str = textwrap.dedent("""\
+
+
+ ]>
+
+
+
+
+ """)
+
+ # Create the problem
+ problem = new_loncapa_problem(xml_str)
+
+ # Render the HTML
+ the_html = problem.get_html()
+ self.assertRegexpMatches(the_html, r"\s+
")
+
def _create_test_file(self, path, content_str):
test_fp = self.system.filestore.open(path, "w")
test_fp.write(content_str)