wassname/catalyst
1
0
Fork 0
mirror of https://github.com/wassname/catalyst.git synced 2026-08-11 11:16:15 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #924 from quantopian/dataset-subclassing

Browse Source 
ENH: Make datasets have subclass relationships
This commit is contained in:
Joe Jevnik
2015-12-29 11:43:55 -05:00
parent 1f137d4dd8 54c58d1205
commit 7a6ba4f249
6 changed files with 125 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/pipeline/test_blaze.py
+1 -1
View File
@@ -90,7 +90,7 @@ class BlazeToPipelineTestCase(TestCase):
self.assertEqual(ds.__name__, name)
self.assertTrue(issubclass(ds, DataSet))
self.assertEqual(
{c.name: c.dtype for c in ds._columns},
{c.name: c.dtype for c in ds.columns},
{'sid': np.int64, 'value': np.float64},
)
tests/pipeline/test_term.py
+71
View File
@@ -28,6 +28,14 @@ class SomeDataSet(DataSet):
buzz = Column(float64_dtype)
class SubDataSet(SomeDataSet):
pass
class SubDataSetNewCol(SomeDataSet):
qux = Column(float64_dtype)
class SomeFactor(Factor):
dtype = float64_dtype
window_length = 5
@@ -321,3 +329,66 @@ class ObjectIdentityTestCase(TestCase):
with self.assertRaises(InvalidDType):
SomeFactor(dtype=1)
class SubDataSetTestCase(TestCase):
def test_subdataset(self):
some_dataset_map = {
column.name: column for column in SomeDataSet.columns
}
sub_dataset_map = {
column.name: column for column in SubDataSet.columns
}
self.assertEqual(
{column.name for column in SomeDataSet.columns},
{column.name for column in SubDataSet.columns},
)
for k, some_dataset_column in some_dataset_map.items():
sub_dataset_column = sub_dataset_map[k]
self.assertIsNot(
some_dataset_column,
sub_dataset_column,
'subclass column %r should not have the same identity as'
' the parent' % k,
)
self.assertEqual(
some_dataset_column.dtype,
sub_dataset_column.dtype,
'subclass column %r should have the same dtype as the parent' %
k,
)
def test_add_column(self):
some_dataset_map = {
column.name: column for column in SomeDataSet.columns
}
sub_dataset_new_col_map = {
column.name: column for column in SubDataSetNewCol.columns
}
sub_col_names = {column.name for column in SubDataSetNewCol.columns}
# check our extra col
self.assertIn('qux', sub_col_names)
self.assertEqual(
sub_dataset_new_col_map['qux'].dtype,
float64_dtype,
)
self.assertEqual(
{column.name for column in SomeDataSet.columns},
sub_col_names - {'qux'},
)
for k, some_dataset_column in some_dataset_map.items():
sub_dataset_column = sub_dataset_new_col_map[k]
self.assertIsNot(
some_dataset_column,
sub_dataset_column,
'subclass column %r should not have the same identity as'
' the parent' % k,
)
self.assertEqual(
some_dataset_column.dtype,
sub_dataset_column.dtype,
'subclass column %r should have the same dtype as the parent' %
k,
)