Should not raise an exception during parsing of XML box. #184

glymur/jp2box.py

@@ -2631,19 +2631,22 @@ class XMLBox(Jp2kBox):
         read_buffer = fptr.read(num_bytes)
         try:
             text = read_buffer.decode('utf-8')
-        except UnicodeDecodeError as ude:
+        except UnicodeDecodeError as err:
             # Possibly bad string of bytes to begin with.
             # Try to search for <?xml and go from there.
             decl_start = read_buffer.find(b'<?xml')
-            if decl_start > -1:
-                text = read_buffer[decl_start:].decode('utf-8')
-            else:
-                raise
+            if decl_start <= -1:
+                msg = 'A problem was encountered while parsing an XML box:'
+                msg += '\n\n\t"{0}"\n\nNo XML was retrieved.'
+                warnings.warn(msg.format(str(err)))
+                return XMLBox(xml=None, length=length, offset=offset)
+
+            text = read_buffer[decl_start:].decode('utf-8')
 
             # Let the user know that the XML box was problematic.
             msg = 'A UnicodeDecodeError was encountered parsing an XML box at '
             msg += 'byte position {0} ({1}), but the XML was still recovered.'
-            msg = msg.format(offset, ude.reason)
+            msg = msg.format(offset, err.reason)
             warnings.warn(msg, UserWarning)
 
         # Strip out any trailing nulls, as they can foul up XML parsing.

glymur/test/test_codestream.py

@@ -40,7 +40,7 @@ class TestCodestream(unittest.TestCase):
                 Jp2k(filename)
         else:
             with self.assertWarns(UserWarning):
-                Jp2k(jfile)
+                Jp2k(filename)
 
 
     @unittest.skipIf(OPJ_DATA_ROOT is None,

glymur/test/test_jp2box_xml.py

@@ -39,6 +39,7 @@ from glymur.jp2box import ColourSpecificationBox, ContiguousCodestreamBox
 from glymur.jp2box import FileTypeBox, ImageHeaderBox, JP2HeaderBox
 from glymur.jp2box import JPEG2000SignatureBox
 
+from .fixtures import OPJ_DATA_ROOT, opj_data_file
 
 @unittest.skipIf(os.name == "nt", "Temporary file issue on window.")
 class TestXML(unittest.TestCase):
@@ -93,6 +94,22 @@ class TestXML(unittest.TestCase):
     def tearDown(self):
         os.unlink(self.xmlfile)
 
+    def test_invalid_utf8(self):
+        """Bad byte sequence that cannot be parsed."""
+        filename = opj_data_file(os.path.join('input',
+                                              'nonregression',
+                                              '26ccf3651020967f7778238ef5af08af.SIGFPE.d25.527.jp2'))
+        if sys.hexversion < 0x03000000:
+            with warnings.catch_warnings():
+                warnings.simplefilter("ignore")
+                jp2 = Jp2k(filename)
+        else:
+            with self.assertWarns(UserWarning):
+                jp2 = Jp2k(filename)
+
+        self.assertIsNone(jp2.box[3].box[1].box[1].xml)
+            
+
     def test_negative_file_and_xml(self):
         """The XML should come from only one source."""
         xml_object = ET.parse(self.xmlfile)