Data Management Examples

These recipes use a products collection. See the five-minute example for a schema and starter data.

Inspect and page through data

curl "http://localhost:9200/collections?offset=0&limit=20&sort_by=name:asc"
curl "http://localhost:9200/collections/products/documents?offset=0&limit=25"
curl "http://localhost:9200/collections/products/documents/product-42?include_created_at=true"

Create a document with a known ID

Stable IDs make retries safe and simplify updates:

curl --fail-with-body -X POST \
  http://localhost:9200/collections/products/documents \
  -H "Content-Type: application/json" \
  -d '{
    "id": "product-42",
    "title": "USB-C Dock",
    "description": "Dock with Ethernet and two display outputs",
    "category": "accessories",
    "price": 89.00,
    "in_stock": true
  }'

Omit id when a generated ID is acceptable.

Apply a partial update

Only supplied fields change:

curl --fail-with-body -X PUT \
  http://localhost:9200/collections/products/documents/product-42 \
  -H "Content-Type: application/json" \
  -d '{"price":79.00,"in_stock":false}'

The update route has upsert behavior. Check existence first if accidentally creating a missing ID would be undesirable.

Import JSON in bulk

For small batches, send a JSON array:

curl --fail-with-body -X POST \
  http://localhost:9200/collections/products/documents/import \
  -H "Content-Type: application/json" \
  -d '[
    {"id":"cable-1","title":"USB-C Cable","price":12.50},
    {"id":"cable-2","title":"Thunderbolt Cable","price":35.00}
  ]'

For large files, stream newline-delimited JSON without loading it into shell memory:

products.ndjson

{"id":"stand-1","title":"Laptop Stand","price":45.00}
{"id":"sleeve-1","title":"Laptop Sleeve","price":24.00}

curl --fail-with-body -X POST \
  http://localhost:9200/collections/products/documents/import \
  -H "Content-Type: application/x-ndjson" \
  --data-binary @products.ndjson

Inspect the import response rather than assuming every row succeeded.

Update documents matching a query

curl --fail-with-body -X POST \
  http://localhost:9200/collections/products/documents/_update_by_query \
  -H "Content-Type: application/json" \
  -d '{
    "filter_by": "category:accessories && price:<20",
    "set": {"in_stock": false}
  }'

Use the exact request shape documented in Update Documents by Query for your server version.

Export selected data

curl --fail-with-body -X POST \
  http://localhost:9200/collections/products/documents/export \
  -H "Content-Type: application/json" \
  -d '{
    "filter_by": "in_stock:true",
    "export_fields": "id,title,price,category"
  }' \
  --output products-export.json

Delete narrowly, then broadly

Delete one document:

curl --fail-with-body -X DELETE \
  http://localhost:9200/collections/products/documents/product-42

Before deleting by filter, run the same filter as a search and inspect found:

curl --fail-with-body -X POST \
  http://localhost:9200/collections/products/documents/search \
  -H "Content-Type: application/json" \
  -d '{"q":"*","filter_by":"category:discontinued","limit":1}'

curl --fail-with-body -X DELETE \
  "http://localhost:9200/collections/products/documents?filter_by=category:discontinued"

Replace an index without changing application code

Create and populate products_v2, then point the stable alias at it:

curl --fail-with-body -X PUT \
  http://localhost:9200/aliases/products_live \
  -H "Content-Type: application/json" \
  -d '{"collection_name":"products_v2"}'

curl --fail-with-body http://localhost:9200/aliases/products_live

Applications can query products_live while new index versions use new physical collection names. See Aliases for the full reindexing pattern.