function read_csv

read_csv(path: str | os.PathLike[str], *, schema: pa.Schema | None = None, columns: Sequence[str] | None = None, names: Sequence[str] | None = None, delimiter: str = ',', header: bool = True, quotechar: str = '"', doublequote: bool = True, escapechar: str = '', skiprows: int = 0, skip_blank_lines: bool = False, null_values: Sequence[str] = ('',), block_size: int = 1048576) -> pa.Table

Source: packages/irx/src/irx/table_io.py:380

Read all CSV rows with strict parsing before optional projection.

The eager result must fit in memory. Inference freezes after the first block; schema controls parsing. Names require header=False and no schema. Quoted empty strings remain strings, unlike default unquoted empty nulls.

Parameters

• path — Type: str | os.PathLike[str]

• schema — Type: pa.Schema | None

• columns — Type: Sequence[str] | None

• names — Type: Sequence[str] | None

• delimiter — Type: str

• header — Type: bool

• quotechar — Type: str

• doublequote — Type: bool

• escapechar — Type: str

• skiprows — Type: int

• skip_blank_lines — Type: bool

• null_values — Type: Sequence[str]

• block_size — Type: int

Returns

Type: pa.Table