Infer a schema from sample data

infer_schema reads a CSV sample and returns the columns, Arrow types (with nullability), and domain candidates - low-cardinality string columns worth locking down. Always infer before importing so you catch type surprises while they're cheap to fix.

Infer

infer_schema(csv_content="product_id,name,category,price\nP001,Widget,Electronics,29.99\n...")

Returns: row count, column names, Arrow types + nullability, and a list of domain candidates (low-cardinality string columns, with their distinct values). The delimiter is auto-detected for parsing.

Review before committing

Common adjustments after looking at the output:

Then import

Feed the reviewed schema to import_csv - it creates the table and inserts the rows in one step. Set domains on the flagged candidates (Govern with domains), then validate.

Next

• Import a CSV · Inspect the live schema.