from_arrow conversion ignores array slicing

[bmschmidt]

Thank you for your work on this excellent project.

Are you using Python or Rust?

Python

What version of polars are you using?

0.7.16

What operating system are you using polars on?

OS X 10.14.3

Describe your bug.

When using pl.from_arrow on python tables with utf8() columns, something is happening that doesn't respect existing slices of arrays.

What are the steps to reproduce the behavior?

import pyarrow as pa
import polars as pl

letters = pa.array(["A", "B", "C"])
tab = pa.table({
    'leading': letters[:-1],
    'lagging': letters[1:]
})
pl.from_arrow(tab)

Gives:

leading	lagging
str	str
"A"	"A"
"B"	"B"

While tab.to_pandas() gives, correctly

leading | lagging
-- | -- | --
A | B
B | C

What is the expected behavior?

The to_pandas() output is correct.

Note: this behavior does not appear if the pyarrow column is an integer rather a string, i.e. the following code works as expected.

import pyarrow as pa
import numpy as np
import polars as pl

numbers = pa.array(np.arange(3))
tab = pa.table({
    'lagging': numbers[1:],
    'leading': numbers[:-1]
})
tab.to_pandas()
pl.from_arrow(tab)

[ritchie46]

Thanks for your issue report. I will look into it.

[ritchie46]

apache/arrow-rs#335

[ritchie46]

Is merged upstream