> ## 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

Both Polars readers support snapshots via the following functions:

## `read_snapshots`

The `read_snapshots` function reads a snapshot file for a given stream and returns it as either a Polars DataFrame or Lazyframe.

* **Parameters**:
  * `stream`: The name of the stream to read the snapshot for. For example, 'contacts'.
  * `snapshot_dir`: The path to the directory where snapshot files are stored.

* **Example**:

  ```python theme={null}
  # Example usage of read_snapshots
  reader = PolarsReader()
  stream_name = 'contacts'
  snapshot_directory = '/path/to/snapshots'

  # Read snapshot for the 'contacts' stream
  snapshot_df = reader.read_snapshots(stream=stream_name, snapshot_dir=snapshot_directory)

  if snapshot_df is not None:
      print("Snapshot DataFrame:")
      print(snapshot_df)
  else:
      print("No snapshot available.")
  ```

## `snapshot_records`

The `snapshot_records` function updates a snapshot file with new data for a given stream and returns a merged Polars DataFrame or Lazyframe.

* **Parameters**:
  * `stream_data`: DataFrame (or Lazyframe) containing the new data to be merged.
  * `stream`: The name of the stream for the snapshots, e.g., 'contacts'.
  * `snapshot_dir`: Path to the directory where snapshot files are stored.
  * `pk`: Primary key(s) to use when merging snapshot; can be a string or a list of strings.
  * `just_new`: If True, returns just the new data, otherwise returns all merged data.
  * `use_csv`: If True, saves and reads snapshots in CSV format instead of Parquet.
  * `overwrite`: If True, overwrites existing snapshot files instead of merging.

* **Example**:

  ```python theme={null}
  import gluestick as gs

  # Process only new records
  reader = gs.PolarsReader()
  snapshot_directory = '/path/to/snapshots'

  for stream in eval(str(reader)):
      # Get new data
      stream_data = reader.get(stream)

      # Update snapshot and get merged data
      merged_df = reader.snapshot_records(
          stream_data=stream_data,
          stream=stream,
          snapshot_dir=snapshot_directory,
          pk=reader.get_pk(stream),
          just_new=False,
          use_csv=False,
          overwrite=False
      )

      gs.to_export(df, stream, "./etl-output", keys=[reader.get_pk(stream)])
  ```