python-unidiff does not gracefully handle encoding

[wontonst]

Here is a snippet from the diff file

+\373e(attlist (("ID" ID IMPLIED) ("LANG" CDATA IMPLIED) ("REMAP" CDATA IMPLIED) ("XREFL\
ABEL" CDATA IMPLIED) ("REVISIONFLAG" (name-token-group ("CHANGED" "ADDED" "DELETED" "OFF\
")) IMPLIED) ("ARCH" CDATA IMPLIED) ("CONFORMANCE" NMTOKENS IMPLIED) ("OS" CDATA IMPLIED\
) ("REVISION" CDATA IMPLIED) ("USERLEVEL" CDATA IMPLIED) ("VENDOR" CDATA IMPLIED) ("ROLE\
" CDATA IMPLIED)))
+^@\200^B^@9^]^A\246^A\206^A^X^A\247^A^Y^A\373S^A\373E^A4^A^U^A5^A\264^A\304^A\263^A\344\
^A6^A\315^A^O^A\340^A<^A\373
+^A\373,^A\373#^A\373T^AJ^A^T^A\373F^A\316^A\356^A\207^A        ^A\373Y^A\221^A_^A`^AY^A\
\373^\^A\357^Af^A\262^A\373^]^A\373c^A\373N^A\234^A=^Ab^A*^A\373d^A\323^A\337^A^N^A\373&\
^A^C^A\317^A\3739^A\245^A!^A9!^A\245^A\3739^A\317^A^C^A\373&^A^N^A\337^A\323^A\373d^A*^A\
b^A=^A\234^A\373N^A\373c^A\373^]^A\262^Af^A\357^A\373^\^AY^A`^A_^A\221^A\373Y^A ^A\207^A\
\356^A\316^A\373F^A^T^AJ^A\373T^A\373#^A\373,^A\373

This is the error I get.

  File "/tmp/cov_venv/local/lib/python2.7/site-packages/unidiff/patch.py", line 195, in _parse_hunk
    raise UnidiffParseError('Hunk diff line {}: expected: {}'.format(diff_line_no, line))
unidiff.errors.UnidiffParseError: Hunk diff line 4068036: expected: SE456<

Line 4068036 is the last line in the diff input.