Method read

pandas/io/parsers/arrow_parser_wrapper.py:268–328

Reads the contents of a CSV file into a DataFrame and processes it according to the kwargs passed in the constructor. Returns ------- DataFrame The DataFrame created from the CSV file.

(self)

            )

    def read(self) -> DataFrame:
        """
        Reads the contents of a CSV file into a DataFrame and
        processes it according to the kwargs passed in the
        constructor.

        Returns
        -------
        DataFrame
            The DataFrame created from the CSV file.
        """
        pa = import_optional_dependency("pyarrow")
        pyarrow_csv = import_optional_dependency("pyarrow.csv")
        self._get_pyarrow_options()
        convert_options = self._get_convert_options()

        try:
            table = pyarrow_csv.read_csv(
                self.src,
                read_options=pyarrow_csv.ReadOptions(**self.read_options),
                parse_options=pyarrow_csv.ParseOptions(**self.parse_options),
                convert_options=convert_options,
            )
        except pa.ArrowInvalid as e:
            raise ParserError(e) from e

        dtype_backend = self.kwds["dtype_backend"]

        # Convert all pa.null() cols -> float64 (non nullable)
        # else Int64 (nullable case, see below)
        if dtype_backend is lib.no_default:
            new_schema = table.schema
            new_type = pa.float64()
            for i, arrow_type in enumerate(table.schema.types):
                if pa.types.is_null(arrow_type):
                    new_schema = new_schema.set(
                        i, new_schema.field(i).with_type(new_type)
                    )

            table = table.cast(new_schema)

        multi_index_named = self._adjust_column_names(table)

        with warnings.catch_warnings():
            warnings.filterwarnings(
                "ignore",
                "make_block is deprecated",
                Pandas4Warning,
            )
            frame = arrow_table_to_pandas(
                table,
                dtype_backend=dtype_backend,
                null_to_int64=True,
                dtype=self.dtype,
                names=self.names,
            )

        if self.header is None: