import dask
import dask.dataframe as dd
import pandas as pd
import numpy as np

# Generate test parquet data
n_partitions = 10
data = {
    'x': np.random.randn(1000),
    'y': np.random.randn(1000),
}

df = pd.DataFrame(data)
ddf = dd.from_pandas(df, npartitions=n_partitions)
ddf.to_parquet('test.parquet')

# Read back the data
test = dd.read_parquet("test.parquet")

# Partition structure without column projection
print(test.map_partitions(len).compute())

# Partition structure WITH column projection
print(test['x'].map_partitions(len).compute())

# Partition structure WITH column projection, "tune" optimization disabled
dask.config.set({"optimization.tune.active": False})
print(test['x'].map_partitions(len).compute())