Fix assertrepr for mojibake

If the compared text was in bytes and not actually valid text
(i.e. could not be encoded to text/unicode using the default encoding)
then the assertrepr would fail with an EncodingError.  This ensures
that the internal string is always valid unicode, converting any bytes
safely to valid unicode.  This is done using repr() which then needs
post-processing to fix the encompassing quotes and un-escape newlines.

This fixes issue 429.

_pytest/assertion/util.py

@@ -162,12 +162,18 @@ def assertrepr_compare(config, op, left, right):
 
 
 def _diff_text(left, right, verbose=False):
-    """Return the explanation for the diff between text
+    """Return the explanation for the diff between text or bytes
 
     Unless --verbose is used this will skip leading and trailing
     characters which are identical to keep the diff minimal.
+
+    If the input are bytes they will be safely converted to text.
     """
     explanation = []
+    if isinstance(left, py.builtin.bytes):
+        left = u(repr(left)[1:-1]).replace(r'\n', '\n')
+    if isinstance(right, py.builtin.bytes):
+        right = u(repr(right)[1:-1]).replace(r'\n', '\n')
     if not verbose:
         i = 0  # just in case left or right has zero length
         for i in range(min(len(left), len(right))):

testing/test_assertion.py

@@ -185,6 +185,19 @@ class TestAssert_reprcompare:
         assert expl[1] == py.builtin._totext('- £€', 'utf-8')
         assert expl[2] == py.builtin._totext('+ £', 'utf-8')
 
+    def test_mojibake(self):
+        # issue 429
+        left = 'e'
+        right = '\xc3\xa9'
+        if not isinstance(left, py.builtin.bytes):
+            left = py.builtin.bytes(left, 'utf-8')
+            right = py.builtin.bytes(right, 'utf-8')
+        expl = callequal(left, right)
+        for line in expl:
+            assert isinstance(line, py.builtin.text)
+        msg = py.builtin._totext('\n').join(expl)
+        assert msg
+
 
 def test_python25_compile_issue257(testdir):
     testdir.makepyfile("""