Fix unicode bug under Py2.

CHANGELOG.rst

@@ -1,7 +1,7 @@
 1.5.2
 -----
 
-
+* Improved handling of non-ascii encoded CSV files under Python 2.
 
 1.5.1 - December 23, 2016
 -------------------------

agate/table/from_csv.py

@@ -17,15 +17,17 @@ def from_csv(cls, path, column_names=None, column_types=None, row_names=None, sk
 
     :param path:
         Filepath or file-like object from which to read CSV data. If a file-like
-        object is specified, it must be seekable.
+        object is specified, it must be seekable. If using Python 2, the file
+        should be opened in binary mode (`rb`).
     :param column_names:
         See :meth:`.Table.__init__`.
     :param column_types:
         See :meth:`.Table.__init__`.
     :param row_names:
         See :meth:`.Table.__init__`.
     :param skip_lines:
-        The number of lines to skip from the top of the file.
+        The number of lines to skip from the top of the file. Note that skip
+        lines will not work with
     :param header:
         If :code:`True`, the first row of the CSV is assumed to contain column
         names. If :code:`header` and :code:`column_names` are both specified
@@ -46,7 +48,11 @@ def from_csv(cls, path, column_names=None, column_types=None, row_names=None, sk
     if hasattr(path, 'read'):
         f = path
     else:
-        f = io.open(path, encoding=encoding)
+        if six.PY2:
+            f = open(path, 'Urb')
+        else:
+            f = io.open(path, encoding=encoding)
+
         close = True
 
     if isinstance(skip_lines, int):
@@ -63,10 +69,10 @@ def from_csv(cls, path, column_names=None, column_types=None, row_names=None, sk
     elif sniff_limit > 0:
         kwargs['dialect'] = csv.Sniffer().sniff(f.read(sniff_limit))
 
-    f.seek(start)
-
     if six.PY2:
-        f = six.StringIO(f.read().encode('utf-8'))
+        kwargs['encoding'] = encoding
+
+    f.seek(start)
 
     reader = csv.reader(f, header=header, **kwargs)
 

tests/test_table/test_from_csv.py

@@ -3,6 +3,8 @@
 
 import io
 
+import six
+
 from agate import Table
 from agate.testcase import AgateTestCase
 from agate.data_types import *
@@ -55,8 +57,13 @@ def test_from_csv_cr(self):
     def test_from_csv_file_like_object(self):
         table1 = Table(self.rows, self.column_names, self.column_types)
 
-        with io.open('examples/test.csv', encoding='utf-8') as f:
-            table2 = Table.from_csv(f)
+        if six.PY2:
+            f = open('examples/test.csv', 'rb')
+        else:
+            f = io.open('examples/test.csv', encoding='utf-8')
+
+        table2 = Table.from_csv(f)
+        f.close()
 
         self.assertColumnNames(table2, table1.column_names)
         self.assertColumnTypes(table2, [Number, Text, Boolean, Date, DateTime, TimeDelta])