BUG: Fix filter method so that accepts byte and unicode

Author: Licht-T

pandas/core/generic.py

@@ -50,8 +50,9 @@
 from pandas.tseries.frequencies import to_offset
 from pandas import compat
 from pandas.compat.numpy import function as nv
-from pandas.compat import (map, zip, lzip, lrange, string_types,
-                           isidentifier, set_function_name, cPickle as pkl)
+from pandas.compat import (map, zip, lzip, lrange, string_types, binary_type,
+                           bytes_to_str, isidentifier, set_function_name,
+                           cPickle as pkl)
 from pandas.core.ops import _align_method_FRAME
 import pandas.core.nanops as nanops
 from pandas.util._decorators import (Appender, Substitution,
@@ -3218,14 +3219,22 @@ def filter(self, items=None, like=None, regex=None, axis=None):
                 **{name: [r for r in items if r in labels]})
         elif like:
             def f(x):
-                if not isinstance(x, string_types):
+                if isinstance(x, binary_type):
+                    x = bytes_to_str(x)
+                elif not isinstance(x, string_types):
                     x = str(x)
                 return like in x
             values = labels.map(f)
             return self.loc(axis=axis)[values]
         elif regex:
+            def f(x):
+                if isinstance(x, binary_type):
+                    x = bytes_to_str(x)
+                elif not isinstance(x, string_types):
+                    x = str(x)
+                return matcher.search(x) is not None
             matcher = re.compile(regex)
-            values = labels.map(lambda x: matcher.search(str(x)) is not None)
+            values = labels.map(f)
             return self.loc(axis=axis)[values]
         else:
             raise TypeError('Must pass either `items`, `like`, or `regex`')