# orca.warehouse

> A module of Orca's API.

A dataset's current snapshot, read from its table's metadata file without iceberg, for `<binary> sense`.

A dataset `ns.table` is the filesystem catalog's table at
`<ORCA_WAREHOUSE>/ns/table`, as in `orca.tables`. Its current metadata file
is the one `metadata/version-hint.text` names, or else the highest version
under `metadata/`, ties broken by `last-updated-ms`, as iceberg's catalog
finds it. Of that file, `current_snapshot` reads the current snapshot's ID
and two properties of its summary: whether an orca step made it
(`orca.run`), and the partition keys it holds (`orca.partitions`), which a
writer that isn't orca can set too.

The warehouse is a directory or a location in object storage
(`s3://bucket/warehouse`), listed and read through objectstore's
`FileIOResolver`, from the `AWS_*` environment.

orca's pipelines import iceberg only if they use tables, and every binary
answers `sense`, so this reads the JSON itself: just enough of it to walk
objects and arrays to the fields it needs.

## Structs

### `Sensed`

```mojo
struct Sensed
```

A dataset's current snapshot, as `sense` prints it.

Implements: `Copyable`, `Movable`

Fields:

- `dataset: String`
- `snapshot: Int64`: -1 when the table doesn't exist or has no snapshot.
- `by_orca: Bool`: Whether an orca step committed the snapshot.
- `partitions: String`: The snapshot summary's `orca.partitions`, or empty.

## Functions

### `current_snapshot`

```mojo
def current_snapshot(warehouse: String, dataset: String) -> Sensed
```

The dataset's current snapshot in the warehouse at `warehouse`.

### `current_of`

```mojo
def current_of(dataset: String, text: String) -> Sensed
```

The current snapshot a table metadata document names, and its summary's orca properties.