Documents

Documents are plain, flat JSON objects — there's no envelope or wrapper around them. id is implicit: every document has one, but it's never declared as a schema field, and field names starting with _ are reserved.

Indexing

POST /collections/{name}/documents accepts either a single document object or a JSON array of documents, indexed as one batch:

curl -X POST localhost:8108/collections/products/documents \
  -H 'Content-Type: application/json' \
  -d '[
    {"id": "1", "title": "Wireless Mouse", "brand": "Logitech", "price": 2999},
    {"id": "2", "title": "Mechanical Keyboard", "brand": "Razer", "price": 8999}
  ]'

Each document in a batch is indexed independently — one invalid document doesn't fail the rest of the batch. The response reports per-document outcomes:

{
  "num_indexed": 2,
  "num_failed": 0,
  "results": [
    { "success": true, "id": "1", "error": null, "code": null },
    { "success": true, "id": "2", "error": null, "code": null }
  ]
}

Indexing the same id again replaces the existing document.

Fetching and deleting

Deletes take effect immediately for subsequent searches, though the underlying segment data is only physically reclaimed on the next merge — see Persistence.

Full request/response shapes: API Reference → Documents.