polars `read_csv()` to read from string and not from file

is it possible to read from string with pl.read_csv() ? Something like this, which would work :

content = """c1, c2
            A,1,
            B,3
            C,2"""
pl.read_csv(content)

I know of course about this :

pl.DataFrame({"c1":["A", "B", "C"],"c2" :[1,3,2]})

But it is error-prone with long tables and you have to count numbers to know which value to modify.
I also know about dictionnaries but I have more than 2 columns in my real life example.

Context : I used to fread() content with R data.table and it was very useful, especially when you want to convert a column with the help of a join, instead of complicated ifelse() statements

Thanks !

>Solution :

pl.read_csv() accepts IO as source parameter.

source: str | Path | IO[str] | IO[bytes] | bytes

So you can use io.StringIO:

from io import StringIO

content = """
c1,c2
A,1
B,3
C,2
"""

data = StringIO(content)
pl.read_csv(data)

shape: (3, 2)
┌─────┬─────┐
│ c1  ┆ c2  │
│ --- ┆ --- │
│ str ┆ i64 │
╞═════╪═════╡
│ A   ┆ 1   │
│ B   ┆ 3   │
│ C   ┆ 2   │
└─────┴─────┘

As you can see above, you can also pass bytes as source parameter. You can use str.encode() method for that:

content = """
c1,c2
A,1
B,3
C,2
"""

pl.read_csv(content.encode())

shape: (3, 2)
┌─────┬─────┐
│ c1  ┆ c2  │
│ --- ┆ --- │
│ str ┆ i64 │
╞═════╪═════╡
│ A   ┆ 1   │
│ B   ┆ 3   │
│ C   ┆ 2   │
└─────┴─────┘