Writing a Dictionary encoded array with snappy compression leads to pyarrow error

[ritchie46]

When reading a parquet file in pyarrow it leads to an OS-error saying that it was incorrectly snappy encoded.:

tests/test_io.py:26: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
/opt/miniconda3/lib/python3.9/site-packages/polars/io.py:539: in read_parquet
    pa.parquet.read_table(
/opt/miniconda3/lib/python3.9/site-packages/pyarrow/parquet.py:1895: in read_table
    return dataset.read(columns=columns, use_threads=use_threads,
/opt/miniconda3/lib/python3.9/site-packages/pyarrow/parquet.py:1744: in read
    table = self._dataset.to_table(
pyarrow/_dataset.pyx:465: in pyarrow._dataset.Dataset.to_table
    ???
pyarrow/_dataset.pyx:3075: in pyarrow._dataset.Scanner.to_table
    ???
pyarrow/error.pxi:143: in pyarrow.lib.pyarrow_internal_check_status
    ???
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

>   ???
E   OSError: Corrupt snappy compressed data.

[jorgecarleitao]

Thanks! (Migrated to arrow2).

Forgot to compress the dictionary page. I have a refactor of the parquet's API to avoid this, but still needs some work.