Fixed a bug that occurred when pandas DataFrame included a datetime c…

…olumn

CHANGES.md

@@ -9,6 +9,9 @@ v0.15.0 (unreleased)
 v0.14.2 (unreleased)
 --------------------
 
+* Fix bug that caused date/time columns in Excel files to not be
+  read in correctly.
+
 * Fix bug that caused demo VO Table to not be read in correctly with
   recent versions of Numpy and Astropy. [#1911]
 

glue/core/component.py

@@ -10,7 +10,7 @@
 
 
 __all__ = ['Component', 'DerivedComponent', 'CategoricalComponent',
-           'CoordinateComponent']
+           'CoordinateComponent', 'DateTimeComponent']
 
 
 class Component(object):
@@ -43,6 +43,8 @@ def __init__(self, data, units=None):
         # The actual data
         # subclasses may pass non-arrays here as placeholders.
         if isinstance(data, np.ndarray):
+            if data.dtype.kind == 'M':
+                raise TypeError('DateTimeComponent should be used instead of Component for np.datetime64 arrays')
             data = coerce_numeric(data)
             data.setflags(write=False)  # data is read-only
 

glue/core/data_factories/pandas.py

@@ -37,7 +37,7 @@ def panda_process(indf):
                 # play well with np.unique
                 c = CategoricalComponent(column.fillna(''))
         else:
-            c = Component(column.values)
+            c = Component.autotyped(column.values)
 
         # convert header to string - in some cases if the first row contains
         # numbers, these are cast to numerical types, so we want to change that

glue/core/data_factories/tests/test_excel.py

@@ -2,7 +2,8 @@
 
 import os
 
-from numpy.testing import assert_array_equal
+import numpy as np
+from numpy.testing import assert_array_equal, assert_allclose
 
 from glue.core import data_factories as df
 from glue.tests.helpers import requires_xlrd, make_file
@@ -52,3 +53,21 @@ def test_excel_single():
     assert_array_equal(d['a'], ['b', 'c', 'd', 'e'])
 
     assert d.label == 'simple_data:Data2'
+
+
+@requires_xlrd
+def test_excel_datetime():
+
+    from ..excel import panda_read_excel
+
+    d = panda_read_excel(os.path.join(DATA, 'datetime.xlsx'))[0]
+
+    assert d.get_kind('date') == 'datetime'
+    assert d.get_kind('a') == 'numerical'
+    assert d.get_kind('b') == 'numerical'
+
+    expected = np.array(['2019-01-01', '2019-02-01', '2019-03-01', '2019-04-01'], dtype='datetime64[ns]')
+
+    assert_array_equal(d['date'], expected)
+    assert_allclose(d['a'], [61.35, 44.06, 83.02, 66.15])
+    assert_allclose(d['b'], [79.34, 15.66, 84.30, 61.53])