How to delete records?¶

Registry records can be deleted with record.delete(), which will permanently remove them from your database.

When it comes to records of File and Collection, they are “moved into trash” when you first call record.delete().

Trashed records are invisible in the UI and excluded from the query results, see visibility faq.
If a record is already in the trash or permanent=True is passed, calling record.delete() triggers permanent delete.
During permanent deletion of a record, its artifact in storage is also deleted unless it has a semantic key.

Setup¶

!lamin init --storage test-delete

💡 connected lamindb: testuser1/test-delete

import lamindb as ln
import pandas as pd

💡 connected lamindb: testuser1/test-delete

artifact = ln.Artifact.from_df(pd.DataFrame({"a": [1, 2], "b": [3, 4]}), description="mydf")
artifact.save()

❗ no run & transform get linked, consider calling ln.track()

Artifact(updated_at=2024-05-20 08:58:12 UTC, uid='WzjKPWg06AcJmlYlGhTB', suffix='.parquet', accessor='DataFrame', description='mydf', size=2240, hash='pCh1QueKcIO78R19tjOUag', hash_type='md5', visibility=1, key_is_virtual=True, created_by_id=1, storage_id=1)

ln.Artifact.df()

	version	created_at	created_by_id	updated_at	uid	storage_id	key	suffix	accessor	description	size	hash	hash_type	n_objects	n_observations	transform_id	run_id	visibility	key_is_virtual
id
1	None	2024-05-20 08:58:12.147945+00:00	1	2024-05-20 08:58:12.148019+00:00	WzjKPWg06AcJmlYlGhTB	1	None	.parquet	DataFrame	mydf	2240	pCh1QueKcIO78R19tjOUag	md5	None	None	None	None	1	True

Trash an artifact¶

artifact.delete()

❗ moved artifact to trash (visibility = -1)

No longer visible:

ln.Artifact.df()

	version	created_at	updated_at	uid	key	suffix	accessor	description	size	hash	hash_type	n_objects	n_observations	visibility	key_is_virtual	created_by_id	storage_id	transform_id	run_id
id

But the artifact still exists in the database, you can find it by not filtering for visibility:

ln.Artifact.filter(visibility=None).df()

	version	created_at	created_by_id	updated_at	uid	storage_id	key	suffix	accessor	description	size	hash	hash_type	n_objects	n_observations	transform_id	run_id	visibility	key_is_virtual
id
1	None	2024-05-20 08:58:12.147945+00:00	1	2024-05-20 08:58:12.195309+00:00	WzjKPWg06AcJmlYlGhTB	1	None	.parquet	DataFrame	mydf	2240	pCh1QueKcIO78R19tjOUag	md5	None	None	None	None	-1	True

You can restore an artifact from trash:

artifact.restore()

ln.Artifact.df()

	version	created_at	created_by_id	updated_at	uid	storage_id	key	suffix	accessor	description	size	hash	hash_type	n_objects	n_observations	transform_id	run_id	visibility	key_is_virtual
id
1	None	2024-05-20 08:58:12.147945+00:00	1	2024-05-20 08:58:12.244902+00:00	WzjKPWg06AcJmlYlGhTB	1	None	.parquet	DataFrame	mydf	2240	pCh1QueKcIO78R19tjOUag	md5	None	None	None	None	1	True

Permanent delete¶

Calling artifact.delete on a trashed artifact triggers a permanent delete dialog. You can pass permanent=True to auto-confirm the deletion.

artifact.delete(permanent=True)

Now its gone in the database:

ln.Artifact.filter(visibility=None).df()

	version	created_at	updated_at	uid	key	suffix	accessor	description	size	hash	hash_type	n_objects	n_observations	visibility	key_is_virtual	created_by_id	storage_id	transform_id	run_id
id