Method test_invalid_row_handler

python/pyarrow/tests/test_csv.py:725–766

(self, pickle_module)

                            convert_options=convert_options)

    def test_invalid_row_handler(self, pickle_module):
        rows = b"a,b\nc\nd,e\nf,g,h\ni,j\n"
        parse_opts = ParseOptions()
        with pytest.raises(
                ValueError,
                match="Expected 2 columns, got 1: c"):
            self.read_bytes(rows, parse_options=parse_opts)

        # Skip requested
        parse_opts.invalid_row_handler = InvalidRowHandler('skip')
        table = self.read_bytes(rows, parse_options=parse_opts)
        assert table.to_pydict() == {
            'a': ["d", "i"],
            'b': ["e", "j"],
        }

        def row_num(x):
            return None if self.use_threads else x
        expected_rows = [
            InvalidRow(2, 1, row_num(2), "c"),
            InvalidRow(2, 3, row_num(4), "f,g,h"),
        ]
        assert parse_opts.invalid_row_handler.rows == expected_rows

        # Error requested
        parse_opts.invalid_row_handler = InvalidRowHandler('error')
        with pytest.raises(
                ValueError,
                match="Expected 2 columns, got 1: c"):
            self.read_bytes(rows, parse_options=parse_opts)
        expected_rows = [InvalidRow(2, 1, row_num(2), "c")]
        assert parse_opts.invalid_row_handler.rows == expected_rows

        # Test ser/de
        parse_opts.invalid_row_handler = InvalidRowHandler('skip')
        parse_opts = pickle_module.loads(pickle_module.dumps(parse_opts))

        table = self.read_bytes(rows, parse_options=parse_opts)
        assert table.to_pydict() == {
            'a': ["d", "i"],
            'b': ["e", "j"],
        }

    def test_chunker_out_of_sync(self):
        # GH-39892: if there are newlines in values, the parser may become