Use encoding when reading column headers

Author: kshedden

pandas/io/sas/sas7bdat.py

@@ -120,7 +120,8 @@
 
 _compression_literals = [_rle_compression, _rdc_compression]
 
-# Incomplete list of encodings
+# Incomplete list of encodings, using SAS nomenclature:
+# http://support.sas.com/documentation/cdl/en/nlsref/61893/HTML/default/viewer.htm#a002607278.htm
 _encoding_names = {29: "latin1", 20: "utf-8", 33: "cyrillic", 60: "wlatin2",
                    61: "wcyrillic", 62: "wlatin1", 90: "ebcdic870"}
 
@@ -526,7 +527,7 @@ def _process_columntext_subheader(self, offset, length):
 
         buf = self._read_bytes(offset, text_block_size)
         self.column_names_strings.append(
-            buf[0:text_block_size].rstrip(b"\x00 ").decode())
+            buf[0:text_block_size].rstrip(b"\x00 ").decode(self.encoding))
 
         if len(self.column_names_strings) == 1:
             column_name = self.column_names_strings[0]

pandas/io/tests/sas/data/test_12659.csv

@@ -0,0 +1,37 @@
+yearmonth,useGpCo,useGpVi,useSpec,useUrge,caseGpCo,caseGpVi,caseSpec,caseUrge,expendGpCo,expendGpVi,expendSpec,expendUrge
+201401,11,12,13,14,15,16,17,18,19,20,21,22
+201402,11,12,13,14,15,16,17,18,19,20,21,22
+201403,11,12,13,14,15,16,17,18,19,20,21,22
+201404,11,12,13,14,15,16,17,18,19,20,21,22
+201405,11,12,13,14,15,16,17,18,19,20,21,22
+201406,11,12,13,14,15,16,17,18,19,20,21,22
+201407,11,12,13,14,15,16,17,18,19,20,21,22
+201408,11,12,13,14,15,16,17,18,19,20,21,22
+201409,11,12,13,14,15,16,17,18,19,20,21,22
+201410,11,12,13,14,15,16,17,18,19,20,21,22
+201411,11,12,13,14,15,16,17,18,19,20,21,22
+201412,11,12,13,14,15,16,17,18,19,20,21,22
+201501,11,12,13,14,15,16,17,18,19,20,21,22
+201502,11,12,13,14,15,16,17,18,19,20,21,22
+201503,11,12,13,14,15,16,17,18,19,20,21,22
+201504,11,12,13,14,15,16,17,18,19,20,21,22
+201505,11,12,13,14,15,16,17,18,19,20,21,22
+201506,11,12,13,14,15,16,17,18,19,20,21,22
+201507,11,12,13,14,15,16,17,18,19,20,21,22
+201508,11,12,13,14,15,16,17,18,19,20,21,22
+201509,11,12,13,14,15,16,17,18,19,20,21,22
+201510,11,12,13,14,15,16,17,18,19,20,21,22
+201511,11,12,13,14,15,16,17,18,19,20,21,22
+201512,11,12,13,14,15,16,17,18,19,20,21,22
+201601,11,12,13,14,15,16,17,18,19,20,21,22
+201602,11,12,13,14,15,16,17,18,19,20,21,22
+201603,11,12,13,14,15,16,17,18,19,20,21,22
+201604,11,12,13,14,15,16,17,18,19,20,21,22
+201605,11,12,13,14,15,16,17,18,19,20,21,22
+201606,11,12,13,14,15,16,17,18,19,20,21,22
+201607,11,12,13,14,15,16,17,18,19,20,21,22
+201608,11,12,13,14,15,16,17,18,19,20,21,22
+201609,11,12,13,14,15,16,17,18,19,20,21,22
+201610,11,12,13,14,15,16,17,18,19,20,21,22
+201611,11,12,13,14,15,16,17,18,19,20,21,22
+201612,11,12,13,14,15,16,17,18,19,20,21,22

pandas/io/tests/sas/data/test_12659.sas7bdat

@@ -73,3 +73,13 @@ def test_productsales():
     vn = ["ACTUAL", "PREDICT", "QUARTER", "YEAR", "MONTH"]
     df0[vn] = df0[vn].astype(np.float64)
     tm.assert_frame_equal(df, df0)
+
+
+def test_12659():
+    dirpath = tm.get_data_path()
+    fname = os.path.join(dirpath, "test_12659.sas7bdat")
+    df = pd.read_sas(fname, encoding='latin1')
+    fname = os.path.join(dirpath, "test_12659.csv")
+    df0 = pd.read_csv(fname)
+    df0 = df0.astype(np.float64)
+    tm.assert_frame_equal(df, df0)