Revert "Fix group_by on load inconsistently fusing data"

This reverts commit 15a84e4.

datacube/api/_api.py

@@ -68,8 +68,8 @@ def _get_descriptor_for_dataset(self, dataset_type, datasets, group_by, geopolyg
         if not geopolygon:
             geopolygon = get_bounds(datasets, dataset_type.grid_spec.crs)
 
-        datasets.sort(key=group_by.grouping_key)
-        groups = [Group(key, list(group)) for key, group in groupby(datasets, group_by.grouping_key)]
+        datasets.sort(key=group_by.group_by_func)
+        groups = [Group(key, list(group)) for key, group in groupby(datasets, group_by.group_by_func)]
 
         dataset_descriptor['result_min'] = tuple()
         dataset_descriptor['result_max'] = tuple()

datacube/api/core.py

@@ -369,9 +369,9 @@ def product_sources(datasets, group_by):
 
         .. seealso:: :meth:`product_observations` :meth:`product_data`
         """
-        dimension, grouping_key, units, sorting_key, reverse_sort = group_by
-        datasets.sort(key=sorting_key, reverse=reverse_sort)
-        groups = [Group(key, tuple(group)) for key, group in groupby(datasets, grouping_key)]
+        dimension, group_func, units = group_by
+        datasets.sort(key=group_func)
+        groups = [Group(key, tuple(group)) for key, group in groupby(datasets, group_func)]
 
         data = numpy.empty(len(groups), dtype=object)
         for index, group in enumerate(groups):

datacube/api/grid_workflow.py

@@ -223,8 +223,8 @@ def tile_sources(observations, group_by):
         """
         tiles = {}
         for cell_index, observation in observations.items():
-            observation['datasets'].sort(key=group_by.sorting_key, reverse=group_by.reversed)
-            groups = [(key, tuple(group)) for key, group in groupby(observation['datasets'], group_by.grouping_key)]
+            observation['datasets'].sort(key=group_by.group_by_func)
+            groups = [(key, tuple(group)) for key, group in groupby(observation['datasets'], group_by.group_by_func)]
 
             for key, datasets in groups:
                 data = numpy.empty(1, dtype=object)

datacube/api/query.py

@@ -33,7 +33,7 @@
 _LOG = logging.getLogger(__name__)
 
 
-GroupBy = collections.namedtuple('GroupBy', ['dimension', 'grouping_key', 'units', 'sorting_key', 'reversed'])
+GroupBy = collections.namedtuple('GroupBy', ['dimension', 'group_by_func', 'units'])
 
 FLOAT_TOLERANCE = 0.0000001  # TODO: For DB query, use some sort of 'contains' query, rather than range overlap.
 SPATIAL_KEYS = ('latitude', 'lat', 'y', 'longitude', 'lon', 'long', 'x')
@@ -171,16 +171,12 @@ def query_geopolygon(geopolygon=None, **kwargs):
 
 def query_group_by(group_by='time', **kwargs):
     time_grouper = GroupBy(dimension='time',
-                           grouping_key=lambda ds: ds.center_time,
-                           units='seconds since 1970-01-01 00:00:00',
-                           sorting_key=lambda ds: ds.center_time,
-                           reversed=True)
+                           group_by_func=lambda ds: ds.center_time,
+                           units='seconds since 1970-01-01 00:00:00')
 
     solar_day_grouper = GroupBy(dimension='time',
-                                grouping_key=solar_day,
-                                units='seconds since 1970-01-01 00:00:00',
-                                sorting_key=lambda ds: ds.center_time,
-                                reversed=True)
+                                group_by_func=solar_day,
+                                units='seconds since 1970-01-01 00:00:00')
 
     group_by_map = {
         None: time_grouper,

integration_tests/test_full_ingestion.py

@@ -173,8 +173,7 @@ def check_open_with_api(index):
 
     geobox = GeoBox(200, 200, Affine(25, 0.0, 1500000, 0.0, -25, -3900000), CRS('EPSG:3577'))
     observations = datacube.product_observations(product='ls5_nbar_albers', geopolygon=geobox.extent)
-    group_by = GroupBy('time', lambda ds: ds.center_time, 'seconds since 1970-01-01 00:00:00',
-                       lambda ds: ds.center_time, True)
+    group_by = GroupBy('time', lambda ds: ds.center_time, 'seconds since 1970-01-01 00:00:00')
     sources = datacube.product_sources(observations, group_by)
     data = datacube.product_data(sources, geobox, input_type.measurements.values())
     assert data.blue.shape == (1, 200, 200)

tests/api/test_core.py

@@ -15,7 +15,7 @@ def group_func(d):
         {'time': datetime.datetime(2016, 2, 1), 'value': 'bar'}
     ]
 
-    group_by = GroupBy(dimension, group_func, units, group_func, reversed=False)
+    group_by = GroupBy(dimension, group_func, units)
     grouped = Datacube.product_sources(datasets, group_by)
 
     assert str(grouped.time.dtype) == 'datetime64[ns]'

tests/api/test_query.py

@@ -86,7 +86,7 @@ def test_convert_descriptor_query_to_search_query_with_groupby():
     }
     query = DescriptorQuery(descriptor_query)
     assert query.group_by
-    assert callable(query.group_by.grouping_key)
+    assert callable(query.group_by.group_by_func)
     assert query.group_by.dimension == 'time'
     assert query.group_by.units == 'seconds since 1970-01-01 00:00:00'