(#1632) fix for unicode chars in seed files

core/dbt/clients/agate_helper.py

@@ -49,4 +49,6 @@ def from_csv(abspath):
  with dbt.compat.open_file(abspath) as fp:
  if fp.read(1) != BOM:
  fp.seek(0)
- return agate.Table.from_csv(fp, column_types=DEFAULT_TYPE_TESTER)
+
+ file_buf = dbt.compat.read_into_buffer(fp)
+ return agate.Table.from_csv(file_buf, column_types=DEFAULT_TYPE_TESTER)

core/dbt/compat.py

@@ -36,11 +36,13 @@
  from SocketServer import TCPServer
  from Queue import PriorityQueue, Empty as QueueEmpty
  from thread import get_ident
+ from StringIO import StringIO
 else:
  from http.server import SimpleHTTPRequestHandler
  from socketserver import TCPServer
  from queue import PriorityQueue, Empty as QueueEmpty
  from threading import get_ident
+ from io import StringIO
 
 
 def to_unicode(s):
@@ -98,6 +100,18 @@ def open_file(path):
  return open(path, encoding='utf-8')
 
 
+def read_into_buffer(fp):
+ buf = StringIO()
+
+ if WHICH_PYTHON == 2:
+ buf.write(fp.read().encode('utf-8'))
+ else:
+ buf.write(fp.read())
+
+ buf.seek(0)
+ return buf
+
+
 if WHICH_PYTHON == 2:
  # In python 2, classmethod and staticmethod do not allow setters, so you
  # can't treat classmethods as first-class objects like you can regular

test/integration/005_simple_seed_test/data-unicode/seed_unicode.csv

@@ -0,0 +1,2 @@
+id
+Uh – Oh

test/integration/005_simple_seed_test/test_simple_seed.py

@@ -197,3 +197,25 @@ def test_simple_seed(self):
  results = self.run_dbt(["seed"])
  self.assertEqual(len(results), 1)
  self.assertTablesEqual("seed_bom", "seed_expected")
+
+
+class TestSimpleSeedWithUnicode(DBTIntegrationTest):
+
+ @property
+ def schema(self):
+ return "simple_seed_005"
+
+ @property
+ def models(self):
+ return "models"
+
+ @property
+ def project_config(self):
+ return {
+ "data-paths": ['data-unicode']
+ }
+
+ @use_profile('postgres')
+ def test_simple_seed(self):
+ results = self.run_dbt(["seed"])
+ self.assertEqual(len(results), 1)