def manual_labels(session: Session, instance: Instance,
project: CategorizationProject) -> Dataset:
"""Get manual labels from a Categorization project.
Args:
instance: Tamr instance containing project
project: Tamr project containing labels
Returns:
Dataset containing manual labels
Raises:
_dataset.NotFound: If no dataset could be found at the specified URL
Ambiguous: If multiple targets match dataset name
"""
unified_dataset = unified.from_project(session=session,
instance=instance,
project=project)
labels_dataset_name = unified_dataset.name + "_manual_categorizations"
datasets_url = URL(instance=instance, path="datasets")
r = session.get(url=str(datasets_url),
params={"filter": f"name=={labels_dataset_name}"})
matches = r.json()
if len(matches) == 0:
raise _dataset.NotFound(str(r.url))
if len(matches) > 1:
raise _dataset.Ambiguous(str(r.url))
dataset_path = matches[0]["relativeId"]
dataset_url = URL(instance=instance, path=dataset_path)
return _dataset._from_url(session=session, url=dataset_url)
def update_unified_dataset(session: Session, project: MasteringProject) -> Operation:
"""Applies changes to the unified dataset and waits for the operation to complete
Args:
project: Tamr Mastering project
"""
unified_dataset = unified.from_project(session, project.url.instance, project)
return unified.apply_changes(session, unified_dataset)
def update_unified_dataset(session: Session,
project: MasteringProject) -> Operation:
"""Apply changes to the unified dataset and wait for the operation to complete
Args:
project: Tamr Mastering project
"""
unified_dataset = unified.from_project(session, project)
op = unified._apply_changes_async(session, unified_dataset)
return operation.wait(session, op)
def manual_labels(session: Session, project: CategorizationProject) -> Dataset:
"""Get manual labels from a Categorization project.
Args:
project: Tamr project containing labels
Returns:
Dataset containing manual labels
Raises:
dataset.NotFound: If no dataset could be found at the specified URL
dataset.Ambiguous: If multiple targets match dataset name
"""
unified_dataset = unified.from_project(session=session, project=project)
labels_dataset_name = unified_dataset.name + "_manual_categorizations"
return _dataset.by_name(
session=session, instance=project.url.instance, name=labels_dataset_name
)
def _update_unified_dataset_async(
session: Session, project: MasteringProject
) -> Operation:
unified_dataset = unified.from_project(session, project)
return unified._apply_changes_async(session, unified_dataset)