Merge pull request #170 from OpenDataServices/cove-838-float-int

[OpenDataServices/cove#838] Convert integer float to int

Author: Bjwebb

flattentool/input.py

@@ -94,6 +94,8 @@ def convert_type(type_string, value, timezone = pytz.timezone('UTC')):
     elif type_string == '':
         if type(value) == datetime.datetime:
             return timezone.localize(value).isoformat()
+        if type(value) == float and int(value) == value:
+            return int(value)
         return value if type(value) in [int] else text_type(value)
     else:
         raise ValueError('Unrecognised type: "{}"'.format(type_string))

flattentool/tests/fixtures/xlsx/integer2.xlsx

@@ -98,8 +98,25 @@ def test_xlsx_input_integer(self):
 
         assert list(xlsxinput.get_sheet_lines('main')) == \
             [{'colA': 1}]
+        if sys.version_info[0] == 2:
+            assert type(list(xlsxinput.get_sheet_lines('main'))[0]['colA']) == long
+        else:
+            assert type(list(xlsxinput.get_sheet_lines('main'))[0]['colA']) == int
         assert xlsxinput.sub_sheet_names == ['main']
 
+    def test_xlsx_input_integer2(self):
+        xlsxinput = XLSXInput(input_name='flattentool/tests/fixtures/xlsx/integer2.xlsx')
+
+        xlsxinput.read_sheets()
+
+        assert list(xlsxinput.get_sheet_lines('Sheet1')) == \
+            [{'activity-status/@code': 2}]
+        # This is a float, but is converted to an int in the unflatten step, see
+        # test_input_SpreadsheetInput_unflatten.py
+        # 'Basic with float'
+        assert type(list(xlsxinput.get_sheet_lines('Sheet1'))[0]['activity-status/@code']) == float
+        assert xlsxinput.sub_sheet_names == ['Sheet1']
+
     def test_xlsx_input_formula(self):
         """ When a forumla is present, we should use the value, rather than the
         formula itself. """

flattentool/tests/test_input_SpreadsheetInput.py

@@ -61,6 +61,24 @@ def inject_root_id(root_id, d):
         [],
         True
     ),
+    (
+        'Basic with float',
+        # 3.0 is converted to 3
+        # This is needed to handle google docs xlsx properly
+        # https://github.com/OpenDataServices/cove/issues/838
+        [{
+            'ROOT_ID': '1',
+            'id': 2,
+            'testA': 3.0
+        }],
+        [{
+                'ROOT_ID': '1',
+                'id': 2,
+                'testA': 3
+        }],
+        [],
+        True
+    ),
     (
         'Basic with zero',
         [{