Try Opteryx

SHOW MANIFEST FOR

The SHOW MANIFEST FOR statement returns the file-level manifest for a dataset — one row per underlying file, with its size, row count, and per-column statistics (min/max bounds, null counts, string lengths). It answers directly from metadata the engine already holds; it never scans the dataset's actual data files.

Syntax

SHOW MANIFEST FOR <table_name>;

Parameters

  • <table_name> — fully qualified as <workspace>.<collection>.<table_name>.

Result Columns

Column Description
file_path Path of the underlying data file
file_format File format, e.g. PARQUET
record_count Row count in the file
file_size_in_bytes File size on disk
uncompressed_size_in_bytes Uncompressed size, when known
column_uncompressed_sizes_in_bytes Per-column uncompressed sizes
null_counts Per-column null counts
min_values / max_values Per-column bounds, positional by field_ids
field_ids Column identifiers the positional stats line up with
min_lengths / max_lengths Per-column string length bounds
min_k_hashes / histogram_counts / char_class_counts Internal sketches used by the query optimizer

Examples

View a Dataset's Manifest

SHOW MANIFEST FOR workspace.collection.large_dataset;

Inspect a Catalog-Backed Dataset

SHOW MANIFEST FOR workspace.collection.orders;

SHOW MANIFEST FOR is not a subquery source — it cannot be wrapped in FROM (...), filtered, or joined. Filter the returned rows client-side if you only need part of the manifest.

Notes

  • Requires ownership. SHOW MANIFEST FOR exposes file paths and storage layout, not just data — it requires the owner role on the dataset, which is stricter than the reader/writer roles that are enough to SELECT from it.
  • Always returns the whole manifest. SHOW statements have no WHERE clause or column list, and the result is not a subquery source, so there is no way to filter or project it at the source at all — filter the rows client-side if you only need part of the manifest.
  • Free to run. The manifest is metadata the engine already holds for query planning; no data files are read to answer this.
  • Works against both local filesystem datasets and catalog-backed (workspace) datasets.

See Also