Documentation for Upsertion Tasks API¶
POST /embeddings/upsertion-tasks/run — create_upsertion_task¶
Description¶
Creates a new task to insert or update vector records in the default vector collection.
Request Parameters¶
task_id(str, optional): Custom ID to ensure idempotent task creation.items(List[DataItem]): The list of vector documents to upsert.
Each DataItem has:
- object_id (str): Unique identifier of the item.
- payload (dict, optional): Structured metadata or content.
- item_info (dict, optional): Internal metadata (e.g. file refs, source).
Request Example¶
{
"task_id": "task-123",
"items": [
{
"object_id": "obj-1",
"payload": {
"text": "deep learning paper"
}
}
]
}
Response Example¶
{
"id": "task-123",
"status": "pending",
"created_at": "2024-05-20T12:00:00Z",
"updated_at": "2024-05-20T12:00:00Z",
"failed_items": []
}
POST /embeddings/upsertion-tasks/categories/run — create_categories_upsertion_task¶
Description¶
Same as /run, but uses the categories vector database instead of the default one.
GET /embeddings/upsertion-tasks/info?task_id=... — get_upsertion_task¶
Description¶
Fetches metadata and current status of an upsertion task.
Query Parameters¶
task_id(str): The task ID to look up.
GET /embeddings/upsertion-tasks/list — list_upsertion_tasks¶
Description¶
Returns a paginated list of upsertion tasks.
Query Parameters¶
offset(int): Number of tasks to skip.limit(int): Max number of tasks to return.status(TaskStatus, optional): Optional filter (pending,processing, etc.).
Response Example¶
[
{
"id": "task-123",
"status": "done",
"created_at": "2024-05-20T12:00:00Z",
"updated_at": "2024-05-20T12:10:00Z",
"failed_items": []
}
]
PUT /embeddings/upsertion-tasks/restart?task_id=... — restart_upsertion_task¶
Description¶
Restarts an upsertion task (sets status back to pending and pushes to queue).
PUT /embeddings/upsertion-tasks/cancel?task_id=... — cancel_upsertion_task¶
Description¶
Cancels an upsertion task by aborting the background job and marking it canceled.