mapreduce/multiproc| Puts bad unicode excpetion handling under test

Author: doublethefish

tests/unittest_checker_similar.py

@@ -511,3 +511,29 @@ def test_get_map_data():
         lines = [line for idx, line in lineset.enumerate_stripped()]
         assert lines_expected == lines
         assert str(expected_lineset) == str(lineset)
+
+
+def test_ammend_stream_bad_unicode():
+    """ Tests that ammend_stream() handles bad unicode errors """
+    sim = similar.SimilarChecker()
+
+    class CustomException(Exception):
+        pass
+
+    class MimicOtherReadError:
+        def readlines(self, *args, **kwargs):
+            raise CustomException()
+
+    raised = False
+    try:
+        sim.append_stream("--unimportant-name--", MimicOtherReadError())
+    except CustomException:
+        raised = True
+    assert raised
+
+    class MimicBadUnicode:
+        def readlines(self, *args, **kwargs):
+            raise UnicodeDecodeError("a", b"a", 0, 0, "a")
+
+    sim.append_stream("--unimportant-name--", MimicBadUnicode())
+    # nothing should happen that stops us getting here and beyond