[Python] How can I use Python Polars to stream the contents of a Parquet file as CSV text to...

Using Python Polars, how can I modify the following script to stream the contents of a Parquet file as CSV text to standard output?

import polars as pl
import sys
pl.scan_parquet("BTCUSDT-trades-2022-01.parquet").sink_csv(sys.stdout)


Python complains that Lazyframe.sink_csv expects a string argument and not a TextIOWrapper:

Traceback (most recent call last):
File "/mnt/storage/Data/Binance/Market Data/Polars/select_btcusdt_aggtrades.py", line 4, in <module>
pl.scan_parquet("../BTCUSDT/2022/BTCUSDT-trades-2022-01.parquet").sink_csv(sys.stdout)
File "/home/derek/.cache/uv/archive-v0/bOaA51uU_dQEF2peOOxqI/lib/python3.12/site-packages/polars/_utils/unstable.py", line 58, in wrapper
return function(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/derek/.cache/uv/archive-v0/bOaA51uU_dQEF2peOOxqI/lib/python3.12/site-packages/polars/lazyframe/frame.py", line 2717, in sink_csv
path=normalize_filepath(path),
^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/derek/.cache/uv/archive-v0/bOaA51uU_dQEF2peOOxqI/lib/python3.12/site-packages/polars/_utils/various.py", line 225, in normalize_filepath
path = os.path.expanduser(path) # noqa: PTH111
^^^^^^^^^^^^^^^^^^^^^^^^
File "<frozen posixpath>", line 259, in expanduser
TypeError: expected str, bytes or os.PathLike object, not TextIOWrapper


(Note that I wish to stream the Parquet file and not read the entire file into a DataFrame in memory because each input file is quite large and there are twelve files for each year, so reading all of the the input data for a year would quickly consume all available memory in my laptop.)

BTCUSDT-trades-2022-01.zip contains the trades in CSV format between Bitcoin and USDT on the Binance cryptocurrency exchange in January 2022. BTCUSDT-trades-2022-01.parquet contains this trade data in Parquet format.

Continue reading...