What is the best way to read a pin from Posit Connect directly into polars?

What is the best way to read a pin from Posit Connect directly into polars?

I have found a few options that work, but none of them are exactly what I am looking for.

(1) Use pin_download

This method is my favourite, but it requires one extra step, I can't get a polars dataframe directly from pin_read.

import pins
import polars as pl

board = pins.board_connect()
paths = board.pin_download("sam.edwardes/vessel_verbose_raw")
pl.read_parquet(paths)

(2) Use pandas

This works, but requires me to have pandas installed. I also assume there is some kind of performance hit b/c you first read into pandas, then into polars?

import pins
import pandas as pd
import polars as pl

board = pins.board_connect()
pl.DataFrame(board.pin_read("sam.edwardes/vessel_verbose_raw"))

(3) Use fsspec

I have not figured out how to implement this yet, but I found this example for duckdb:

https://github.com/rstudio/pins-python/pull/193

Is there a way to do this with polars?

Related issues

• https://github.com/rstudio/pins-python/issues/153
• https://github.com/rstudio/pins-python/issues/32

Suggestions

It would be nice if pins gave me a choice about what DataFrame library is used with pin_read. For example:

import pins
import polars as pl

board = pins.board_connect()
board.pin_read("sam.edwardes/vessel_verbose_raw", df_library='polars')

Start by reading the existing pin_read and pin_download behavior, then compare the fsspec approach in pull request 193 and the related issues 153 and 32. Determine whether direct Posit Connect reads can support polars without pandas, and consider how a DataFrame-library choice would fit the current API; done means a clear supported approach or implementation path.