def __update_joined_datasets(dataset, update):
"""Update any joined datasets."""
if 'add' in update:
new_dframe = update['add']
for direction, other_dataset, on, j_dataset in dataset.joined_datasets:
if 'add' in update:
if direction == 'left':
# only proceed if on in new dframe
if on in new_dframe.columns:
left_dframe = other_dataset.dframe(padded=True)
# only proceed if new on value is in on column in lhs
if len(set(new_dframe[on]).intersection(
set(left_dframe[on]))):
merged_dframe = join_dataset(left_dframe, dataset, on)
j_dataset.replace_observations(merged_dframe)
# TODO is it OK not to propagate the join here?
else:
# if on in new data join with existing data
if on in new_dframe:
new_dframe = join_dataset(new_dframe, other_dataset, on)
calculate_updates(j_dataset, df_to_jsondict(new_dframe),
parent_dataset_id=dataset.dataset_id)
elif 'delete' in update:
j_dataset.delete_observation(update['delete'])
elif 'edit' in update:
j_dataset.update_observation(*update['edit'])
def join(self, other, on):
"""Join with dataset `other` on the passed columns.
:param other: The other dataset to join.
:param on: The column in this and the `other` dataset to join on.
"""
merged_dframe = self.dframe()
if not len(merged_dframe.columns):
# Empty dataset, simulate columns
merged_dframe = self.place_holder_dframe()
merged_dframe = join_dataset(merged_dframe, other, on)
merged_dataset = self.create()
if self.num_rows and other.num_rows:
merged_dataset.save_observations(merged_dframe)
else:
merged_dataset.build_schema(merged_dframe, set_num_columns=True)
merged_dataset.ready()
self.add_joined_dataset(
('right', other.dataset_id, on, merged_dataset.dataset_id))
other.add_joined_dataset(
('left', self.dataset_id, on, merged_dataset.dataset_id))
return merged_dataset
def join(self, other, on):
"""Join with dataset `other` on the passed columns.
:param other: The other dataset to join.
:param on: The column in this and the `other` dataset to join on.
"""
merged_dframe = self.dframe()
if not len(merged_dframe.columns):
# Empty dataset, simulate columns
merged_dframe = self.place_holder_dframe()
merged_dframe = join_dataset(merged_dframe, other, on)
merged_dataset = self.create()
if self.num_rows and other.num_rows:
merged_dataset.save_observations(merged_dframe)
else:
merged_dataset.build_schema(merged_dframe, set_num_columns=True)
merged_dataset.ready()
self.add_joined_dataset(
('right', other.dataset_id, on, merged_dataset.dataset_id))
other.add_joined_dataset(
('left', self.dataset_id, on, merged_dataset.dataset_id))
return merged_dataset
