Merge pull request #73 from datafuselabs/fix/upload-encoding-utf8

fix: encoding uploaded stage file as utf8

Author: hantmac

databend_py/uploader.py

@@ -72,11 +72,11 @@ def _serialize_data(self, data, compress):
         buf = io.StringIO()
         csvwriter = csv.writer(buf, delimiter=',', quoting=csv.QUOTE_MINIMAL)
         csvwriter.writerows(data)
-        output = buf.getvalue()
+        output = buf.getvalue().encode('utf-8')
         if compress:
             buf = io.BytesIO()
             with gzip.GzipFile(fileobj=buf, mode="wb") as gzwriter:
-                gzwriter.write(output.encode('utf-8'))
+                gzwriter.write(output)
             output = buf.getvalue()
         if self._debug:
             print('upload:_serialize_data %s' % (time.time() - start_time))

tests/test_client.py

@@ -198,6 +198,13 @@ def test_null_to_none(self):
         _, data = client.execute("select NULL as test")
         self.assertIsNone(data[0][0])
 
+    def test_special_chars(self):
+        client = Client.from_url(self.databend_url)
+        client.execute("create or replace table test_special_chars (x string)")
+        client.execute("INSERT INTO test_special_chars (x) VALUES", [('ó')])
+        _, data = client.execute("select * from test_special_chars")
+        self.assertEqual(data, [('ó')])
+
     def test_set_query_id_header(self):
         os.environ["ADDITIONAL_HEADERS"] = "X-DATABENDCLOUD-TENANT=TENANT,X-DATABENDCLOUD-WAREHOUSE=WAREHOUSE"
         client = Client.from_url(self.databend_url)