This commit is contained in:
wangjiming
2026-08-18 14:49:34 +08:00
10 changed files with 313 additions and 36 deletions

View File

@@ -1634,7 +1634,24 @@ class PlatformStore:
def create_dataset(self, payload: dict[str, Any]) -> dict[str, Any]:
dataset_id = payload.get("id") or new_id("ds")
name = str(payload.get("name") or "").strip()
if not name:
raise ValueError("dataset name is required")
with self.connect() as conn:
existing = conn.execute(
"SELECT id, deleted_at FROM datasets WHERE name=?",
(name,),
).fetchone()
if existing and not existing.get("deleted_at"):
raise ValueError(f"dataset name already exists: {name}")
# Soft-deleted records remain in the database for audit/history and
# still participate in the legacy unique constraint. Free the name
# while retaining a traceable tombstone before creating the new row.
if existing:
conn.execute(
"UPDATE datasets SET name=? WHERE id=?",
(f"{name}__deleted__{existing['id']}", existing["id"]),
)
conn.execute(
"""
INSERT INTO datasets
@@ -1643,7 +1660,7 @@ class PlatformStore:
""",
(
dataset_id,
payload["name"],
name,
payload.get("type", "train"),
payload.get("storage_type", "local"),
payload.get("source", "upload"),
@@ -2992,6 +3009,27 @@ class PlatformStore:
).fetchone()
return dict(row)
def link_dataset_file_storage_object(self, file_id: str, storage_object_id: str) -> None:
"""Link an uploaded dataset file to its canonical MinIO object."""
with self.connect() as conn:
conn.execute(
"""
UPDATE dataset_files
SET storage_object_id=?
WHERE id=?
""",
(storage_object_id, file_id),
)
row = conn.execute(
"SELECT dataset_id FROM dataset_files WHERE id=?",
(file_id,),
).fetchone()
if row:
conn.execute(
"UPDATE datasets SET storage_type='minio' WHERE id=?",
(row["dataset_id"],),
)
def storage_objects_for_resource(self, resource_type: str, resource_id: str) -> list[dict[str, Any]]:
with self.connect() as conn:
rows = conn.execute(