Schema files

Conversion needs to know what each relation and column name refers to and what Snowflake type it has. A schema file declares them as JSON. Pass it with --schema to convert, function, census and index.

{
  "ORDERS": [
    ["ORDER_ID", "NUMBER(38,0)"],
    ["CUSTOMER_ID", "NUMBER(38,0)"],
    ["TOTAL", "NUMBER(12,2)"],
    ["ORDERED_AT", "TIMESTAMP_NTZ"]
  ]
}

A name the schema does not declare is refused with SQ0004.

Naming relations and columns

Names resolve as Snowflake resolves them. An unquoted name is upper-cased, so FROM orders reads the relation ORDERS. A quoted name keeps its case, so "orders" is a different relation from ORDERS. When a name matches a declared one only ignoring case, the diagnostic says so and suggests the spelling to declare.

Declare a relation as ORDERS, SALES.ORDERS or ANALYTICS.SALES.ORDERS. A qualified name in the SQL, such as analytics.sales.orders in dbt’s compiled models, reads the declared relation whose parts agree with it, and the converted SQL names that relation as declared:

Declared	SQL	Reads
ORDERS	orders, sales.orders, analytics.sales.orders	ORDERS
ANALYTICS.SALES.ORDERS	orders, sales.orders, Analytics.Sales.Orders	ANALYTICS.SALES.ORDERS
ANALYTICS.SALES.ORDERS and ANALYTICS.ARCHIVE.ORDERS	sales.orders	ANALYTICS.SALES.ORDERS
ANALYTICS.SALES.ORDERS and ANALYTICS.ARCHIVE.ORDERS	orders	refused: it matches both

The session’s DATABASE and SCHEMA complete a name written without them, as in Snowflake: with --session SCHEMA=archive, orders reads ANALYTICS.ARCHIVE.ORDERS. See Sessions.

These are refused with SQ0004, naming the relations involved:

• A name matching several declared relations that no session default settles.
• A qualified name whose declared relation a CTE of the same name hides.
• Two differently qualified names in one statement that read the same declared relation. Declare each under its qualified name.

The converted SQL reads a qualified relation from the database you attach to DuckDB under that name, such as ATTACH 'analytics.duckdb' AS analytics.

Columns

Each column is a [name, type] pair, or an object:

Field	Meaning
name	The column name.
type	Its Snowflake type, such as NUMBER(12,2), VARCHAR or VARIANT.
typed_leaves	true when a VARIANT column’s leaves hold typed values. Default false.
offset	For a TIMESTAMP_TZ column whose every value has one offset: +00:00 or UTC.

--confirm-declared-offsets lets strict mode rely on the offsets the schema declares.

Relations and functions

The top level is either a map of relations, as above, or an object with two sections:

{
  "relations": {
    "ORDERS": [["ORDER_ID", "NUMBER(38,0)"], ["TOTAL", "NUMBER(12,2)"]]
  },
  "functions": {
    "NET_TOTAL": {
      "language": "sql",
      "arguments": [["GROSS", "NUMBER(12,2)"]],
      "returns": "NUMBER(12,2)"
    }
  }
}

A function declares its language (sql or python), its arguments as columns, and what it returns. volatility, null_input and vectorized are optional. Unknown fields are rejected.