Support uploading DataFrames with non-ascii texts in Python 2

cartoframes/data/dataset/registry/dataframe_dataset.py

@@ -1,5 +1,4 @@
 import pandas as pd
-from warnings import warn
 from tqdm import tqdm
 
 from carto.exceptions import CartoException, CartoRateLimitException
@@ -131,24 +130,33 @@ def _rows(df, dataframe_columns_info, with_lnglat):
             if dataframe_columns_info.geom_column and col == dataframe_columns_info.geom_column:
                 geom = decode_geometry(val, dataframe_columns_info.enc_type)
                 if geom:
-                    row_data.append('SRID=4326;{}'.format(geom.wkt))
+                    val = 'SRID=4326;{}'.format(geom.wkt)
                 else:
-                    row_data.append('')
-            else:
-                row_data.append('{}'.format(val))
+                    val = ''
+            row_data.append(_encoded(val))
 
         if with_lnglat:
             lng_val = row[with_lnglat[0]]
             lat_val = row[with_lnglat[1]]
             if lng_val and lat_val:
-                row_data.append('SRID=4326;POINT ({lng} {lat})'.format(lng=lng_val, lat=lat_val))
+                val = 'SRID=4326;POINT ({lng} {lat})'.format(lng=lng_val, lat=lat_val)
             else:
-                row_data.append('')
+                val = ''
+            row_data.append(_encoded(val))
+
+        csv_row = _encoded('|').join(row_data)
+        csv_row += _encoded('\n')
 
-        csv_row = '|'.join(row_data)
-        csv_row += '\n'
+        yield csv_row
 
-        yield csv_row.encode()
+
+def _encoded(val):
+    if isinstance(val, type(u'')):
+        return val.encode('utf-8')
+    elif isinstance(val, type(b'')):
+        return val
+    else:
+        return u'{}'.format(val).encode('utf-8')
 
 
 def _is_null(val):

test/data/dataset/test_dataset.py

@@ -14,7 +14,7 @@
 from cartoframes.auth import Credentials
 from cartoframes.data.clients import SQLClient
 from cartoframes.utils.geom_utils import setting_value_exception
-from cartoframes.utils.columns import normalize_name
+from cartoframes.utils.columns import normalize_name, DataframeColumnsInfo
 from cartoframes.utils.utils import load_geojson
 from cartoframes.data import StrategiesRegistry
 from cartoframes.data.dataset.registry.dataframe_dataset import DataFrameDataset, _rows
@@ -906,3 +906,14 @@ def test_rows_null_geom(self):
         rows = _rows(df, dataframe_columns_info, with_lnglat)
 
         self.assertEqual(list(rows), [b'|\n', b'|\n'])
+
+    def test_rows_non_ascii(self):
+        attribute = 'áéí'
+        unicode_attribute = u'áéí'
+        encoded_attribute = unicode_attribute.encode('utf-8')
+        encoded_line = encoded_attribute + '\n'.encode()
+
+        df = pd.DataFrame.from_dict({'test': [attribute, unicode_attribute, encoded_attribute, 'xyz']})
+        columns_info = DataframeColumnsInfo(df)
+        rows = _rows(df, columns_info, None)
+        self.assertEqual(list(rows), [encoded_line, encoded_line, encoded_line, b'xyz\n'])