> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hotglue.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Snapshotting functions

## `readSnapshots` Function

This function reads snapshot data from either Parquet or CSV files for a specified stream.

**Parameters**:

* `stream` (string): The name of the stream to read snapshots for.
* `snapshotDir` (string): The directory containing the snapshot files.
* `options` (object): Optional. CSV read options.

**Returns**:

* A `Polars.DataFrame` containing the snapshot data, or `null` if no snapshot exists.

**Example Usage**:

```typescript theme={null}
const snapshot = readSnapshots('users', '/path/to/snapshot/dir');
if (snapshot) {
  console.log('Snapshot loaded:', snapshot);
} else {
  console.log('No snapshot available.');
}
```

## `snapshotRecords` Function

This function creates or updates snapshot records for a data stream with options for deduplication, type coercion, and more.

**Parameters**:

* `streamData` (Polars.DataFrame or null): The data to snapshot.
* `stream` (string): The name of the stream.
* `snapshotDir` (string): The directory to store snapshot files.
* `pk` (string): Primary key for deduplication. Defaults to "id".
* `justNew` (boolean): Whether to return only new data instead of the full snapshot.
* `useCsv` (boolean): Whether to write snapshots as CSV instead of Parquet.
* `coerceTypes` (boolean): If true, types are coerced to match `streamData` dtypes.
* `localizeDatetimeTypes` (boolean): If true, converts datetime columns to UTC.
* `overwrite` (boolean): If true, overwrites the snapshot entirely instead of merging.
* `options` (object): Additional options for CSV reading.

**Returns**:

* The updated `Polars.DataFrame` of the snapshot, merged according to the primary key specified.

**Example Usage**:

```typescript theme={null}
const df = new Polars.DataFrame({ columns: ['user_id', 'name'] });
const dfWithSnapshottedData = snapshotRecords(df, 'users', '/path/to/snapshot/dir');
console.log('Updated Snapshot:', dfWithSnapshottedData);
```