Updated based on feedback from jreback

Author: tomrod

doc/source/whatsnew/v0.20.0.txt

@@ -271,7 +271,7 @@ Performance Improvements
 - Improved performance of ``pd.wide_to_long()`` (:issue:`14779`)
 - Increased performance of ``pd.factorize()`` by releasing the GIL with ``object`` dtype when inferred as strings (:issue:`14859`)
 
-- When reading buffer object in ``read_sas()`` method without specified format, filepath string is inferred rather than buffer object. Error is now thrown if buffer object is provided without format {sas7bdat|xport} specification.
+- When reading buffer object in ``read_sas()`` method without specified format, filepath string is inferred rather than buffer object.
 
 
 .. _whatsnew_0200.bug_fixes:

pandas/io/sas/sasreader.py

@@ -29,11 +29,11 @@ def read_sas(filepath_or_buffer, format=None, index=None, encoding=None,
     DataFrame if iterator=False and chunksize=None, else SAS7BDATReader
     or XportReader
     """
-
+    from pandas import compat
     if format is None:
-        bufferr = "Format unrecognized. If buffer object, specify format")
-        if type(filesize_or_buffer) != str:
-            raise TypeError(bufferr)
+        buffErr = "Format unrecognized. If buffer object, specify format")
+        if not isinstance(filepath_or_buffer,compat.string_types):
+            raise TypeError(buffErr)
         try:
             fname = filepath_or_buffer.lower()
             if fname.endswith(".xpt"):

pandas/io/tests/sas/test_sas.py

@@ -0,0 +1,5 @@
+def test_sas_buffer_format(self):
+    import StringIO
+    b = StringIO.StringIO("")
+    with self.assertRaises(TypeError):
+        result=pd.read_sas(b)