A method to query the unique values of a partition (name) from a `ParquetDataset`, using `pieces`.

query_unique_value.py

import re
import pyarrow.parquet as pq

def query_unique_value(dataset: pq.ParquetDataset,
                       partition: str
                      ) -> set:
    ''' query the unique values of a given partition name from a give `ParquetDataset`, returns a set.
    
        Parameters
        ----------
        dataset : pyarrow.parquet.ParquetDataset
            The ParquetDataset for the query.
        partition : str
            The name of the partition to query on.

        Returns
        -------
        set
            The unique values of the partition
    '''
    unique_values = set()
    pattern = re.compile(f'.*/{partition}=([^/]*)/')
    for p in dataset.pieces:
        value = re.match(pattern, p.path).group(1)
        unique_values.add(value)
    return unique_values