fix #495: fixed reading 1D arrays with non-string labels

* added test for reading a csv file from a StringIO object
* simplified some test files: test1d.csv, test2d.csv and test3d.csv

Author: gdementen

doc/source/changes/version_0_26_1.rst.inc

@@ -11,3 +11,5 @@ Fixes
 
 * fixed LArray.split_axes using a custom separator and not using sort=True or when the split labels are
   ambiguous with labels from other axes (closes :issue:`485`).
+
+* fixed reading 1D arrays with non-string labels (closes :issue:`495`).

larray/inout/array.py

@@ -234,6 +234,8 @@ def df_aslarray(df, sort_rows=False, sort_columns=False, raw=False, parse_header
     # handle 1D
     if len(df) == 1 and (pd.isnull(df.index.values[0]) or
                              (isinstance(df.index.values[0], basestring) and df.index.values[0].strip() == '')):
+        if parse_header:
+            df.columns = pd.Index([parse(cell) for cell in df.columns.values], name=df.columns.name)
         series = df.iloc[0]
         series.name = df.index.name
         return from_series(series, sort_rows=sort_rows)

larray/tests/test1d.csv

@@ -1,2 +1,2 @@
 time,2007,2010,2013
-,3722,3395,3347
+,0,1,2

larray/tests/test2d.csv

@@ -1,6 +1,3 @@
-age\time,2007,2010,2013
-0,3722,3395,3347
-1,338,316,323
-2,2878,2791,2822
-3,1121,1037,976
-4,4073,4161,4429
+a\b,0,1,2
+0,0,1,2
+1,3,4,5

larray/tests/test3d.csv

@@ -1,11 +1,9 @@
-age,sex\time,2007,2010,2013
-0,F,3722,3395,3347
-0,H,338,316,323
-1,F,2878,2791,2822
-1,H,1121,1037,976
-2,F,4073,4161,4429
-2,H,1561,1463,1467
-3,F,3507,3741,3366
-3,H,2052,2052,2118
-4,F,4807,4868,4852
-4,H,3785,3508,3172
+age,sex\time,2015,2016,2017
+0,F,0.5,1.5,2.5
+0,M,3.5,4.5,5.5
+1,F,6.5,7.5,8.5
+1,M,9.5,10.5,11.5
+2,F,12.5,13.5,14.5
+2,M,15.5,16.5,17.5
+3,F,18.5,19.5,20.5
+3,M,21.5,22.5,23.5

larray/tests/test_array.py

@@ -18,6 +18,7 @@
                     clip, exp, where, X, mean, isnan, round, read_hdf, read_csv, read_eurostat, read_excel,
                     from_lists, from_string, open_excel, from_frame, sequence, nan_equal)
 from larray.core.axis import _to_ticks, _to_key
+from larray.util.misc import StringIO
 
 
 class TestValueStrings(TestCase):
@@ -2455,23 +2456,15 @@ def test_from_string(self):
         assert_array_equal(res, expected)
 
     def test_read_csv(self):
-        la = read_csv(abspath('test1d.csv'))
-        self.assertEqual(la.ndim, 1)
-        self.assertEqual(la.shape, (3,))
-        self.assertEqual(la.axes.names, ['time'])
-        assert_array_equal(la, [3722, 3395, 3347])
+        res = read_csv(abspath('test1d.csv'))
+        assert_array_equal(res, ndrange('time=2007,2010,2013'))
 
-        la = read_csv(abspath('test2d.csv'))
-        self.assertEqual(la.ndim, 2)
-        self.assertEqual(la.shape, (5, 3))
-        self.assertEqual(la.axes.names, ['age', 'time'])
-        assert_array_equal(la[0, :], [3722, 3395, 3347])
+        res = read_csv(abspath('test2d.csv'))
+        assert_array_equal(res, ndrange('a=0,1;b=0,1,2'))
 
-        la = read_csv(abspath('test3d.csv'))
-        self.assertEqual(la.ndim, 3)
-        self.assertEqual(la.shape, (5, 2, 3))
-        self.assertEqual(la.axes.names, ['age', 'sex', 'time'])
-        assert_array_equal(la[0, 'F', :], [3722, 3395, 3347])
+        res = read_csv(abspath('test3d.csv'))
+        expected = ndrange('age=0..3;sex=F,M;time=2015..2017') + 0.5
+        assert_array_equal(res, expected)
 
         la = read_csv(abspath('test5d.csv'))
         self.assertEqual(la.ndim, 5)
@@ -2499,6 +2492,10 @@ def test_read_csv(self):
         assert_array_equal(la[X.arr[1], 0, 'F', X.nat[1], :],
                            [3722, 3395, 3347])
 
+        # test StringIO
+        res = read_csv(StringIO('a,1,2\n,0,1\n'))
+        assert_array_equal(res, ndrange('a=1,2'))
+
     def test_read_eurostat(self):
         la = read_eurostat(abspath('test5d_eurostat.csv'))
         self.assertEqual(la.ndim, 5)