[testsuite] Drop characters that can't be decoded, restoring parity with Py2.

Tests that check the output of `memory find` may trip over
unreadable characters, and in Python 3 this is an error.

llvm-svn: 355612

Author: dcci

lldb/packages/Python/lldbsuite/support/encoded_file.py

@@ -31,6 +31,9 @@ def impl(s):
         # as unicode before attempting to write.
         if isinstance(s, six.binary_type):
             s = s.decode(encoding, "replace")
+        # Filter unreadable characters, Python 3 is stricter than python 2 about them.
+        import re
+        s = re.sub(r'[^\x00-\x7f]',r' ',s)
         return old_write(s)
     return impl