query-dataset

First get the resource UUID from package_show for the dataset slug, then call datastore_search with that resource_id plus optional filters, fields, sort, and limit parameters. Only resources with datastore_active set to true can be queried this way.

Use the datastore_search_sql endpoint with a SQL string where the table name is the resource UUID in double quotes. Column names with accents or spaces must be double-quoted, string values use single quotes, and JOINs across resources are not supported.

Each query returns a maximum of 32,000 rows, with a default limit of 100. To retrieve an entire dataset, paginate with limit and offset parameters until a page returns fewer rows than the limit.

Many datasets store numeric fields as text, so comparisons and aggregations behave unexpectedly. Use CAST("field" AS FLOAT) or CAST("field" AS INTEGER) in the SQL query to convert values before numeric operations.

No, resources without datastore_active cannot be queried through the API. For GeoJSON, Shapefile, PDF, or GBFS resources, extract the download URL from package_show and fetch the file directly with curl.