read_parquet from multiple files

[Gabriel-ROBIN]

What language are you using?

Python 3.8,

What version of polars are you using?

0.13.4

What operating system are you using polars on?

Redhat 7.5

Describe your bug.

Form the doc, we can see that it is possible to read a list of parquet files. I did not make it work.

What are the steps to reproduce the behavior?

Example

import polars as pl

df1 = pl.DataFrame({"bar": [1, 2, 3], "foo": [3, 2, 1]})
df2 = pl.DataFrame({"bar": [4, 5, 6], "foo": [3, 2, 1]})
df1.to_parquet('./abc.parquet')
df2.to_parquet('./def.parquet')

pl.read_parquet([
    './abc.parquet',
    './def.parquet'
])

What is the actual behavior?

Fails with
TypeError: Object does not have a .read() method.

What is the expected behavior?

abc.parquet and def.parquet are concatenate

[ritchie46]

You can only read multiple files at once using glob patters. E.g.

pl.read_parquet(
    './*.parquet',
)

If you want to read without glob patters, you need to call read_parquet separately and concat the DataFrames.

[Gabriel-ROBIN]

Ok, that's what I understood, but in this case there is a typo in the doc string right ? It is said that source could be a List[str]

pl.read_parquet(
    source: Union[str, List[str], pathlib.Path, BinaryIO, _io.BytesIO, bytes],
    columns: Union[List[int], List[str], NoneType] = None,
    ...
) -> polars.internals.frame.DataFrame

[ritchie46]

Ah, check. Thanks for pointing that out.

[ritchie46]

Fixed

[alf239]

Just for the next one bumping into this ticket: it's now added with #13044