Skip to main content
Last updated on

current_snapshot_version function

Returns the version row selected by the enclosing flow's WITH VERSION (...) clause for the snapshot being processed.

This function is available only inside the FROM SNAPSHOT (...) clause of an AUTO CDC ... FROM SNAPSHOT flow.

Syntax​

SQL
current_snapshot_version()

Arguments​

This function takes no arguments.

Returns​

A single-column relation with the same column name and type as the enclosing flow's version query, containing exactly 1 row. The row contains the version selected for the current snapshot-processing iteration.

current_snapshot_version() is evaluated only after the version query returns 1 row. If the version query returns 0 rows, the engine does not run the snapshot query, and it ends processing for the flow during that pipeline update.

Examples​

The following flow uses current_snapshot_version() to read the file selected by the WITH VERSION (...) clause. The version is a single STRUCT column, so its fields are accessed with dot notation:

SQL
CREATE FLOW orders_cdc
  AS AUTO CDC INTO orders
  FROM SNAPSHOT (
    SELECT *
    FROM read_files('/Volumes/catalog/schema/landing/orders/', format => 'json')
    WHERE _metadata.file_path = (SELECT version.path FROM current_snapshot_version())
  )
  WITH VERSION (...)
  KEYS (order_id);

Errors​

current_snapshot_version() is available only inside the FROM SNAPSHOT (...) clause of an AUTO CDC ... FROM SNAPSHOT flow that declares a WITH VERSION (...) clause. Anywhere else, Spark resolves the name as an ordinary table-valued function. Because no such function is defined, it raises UNRESOLVABLE_TABLE_VALUED_FUNCTION (SQLSTATE 42883).

On this page