polars.Expr.bin.reinterpret#

Expr.bin.reinterpret( *, dtype: PolarsDataType | DataTypeExpr, endianness: Endianness = 'little', ) → Expr[source]#

Interpret bytes as another type.

Supported types are numerical or temporal dtypes, or an Array of these dtypes.

Parameters:

dtypePolarsDataType: Which type to interpret binary column into.
endianness{“big”, “little”}, optional: Which endianness to use when interpreting bytes, by default “little”.

Returns:

Expr: Expression of data type dtype. Note that rows of the binary array where the length does not match the size in bytes of the output array (number of items * byte size of item) will become NULL.

Examples

>>> df = pl.DataFrame({"data": [b"\x05\x00\x00\x00", b"\x10\x00\x01\x00"]})
>>> df.with_columns(  
...     bin2int=pl.col("data").bin.reinterpret(
...         dtype=pl.Int32, endianness="little"
...     ),
... )
shape: (2, 2)
┌─────────────────────┬─────────┐
│ data                ┆ bin2int │
│ ---                 ┆ ---     │
│ binary              ┆ i32     │
╞═════════════════════╪═════════╡
│ b"\x05\x00\x00\x00" ┆ 5       │
│ b"\x10\x00\x01\x00" ┆ 65552   │
└─────────────────────┴─────────┘