to_parquet

to_parquet()

to_parquet(input_path: str, output_path: str, compression: str | None = None, row_group_size: typing.SupportsInt | None = None, num_threads: typing.SupportsInt | None = None, error_mode: str | None = None, max_errors: typing.SupportsInt | None = None, comment: str | None = None, skip_empty_rows: bool = True, guess_integer: bool = True, trim_ws: bool = True, escape_backslash: bool = False, decimal_mark: str = ‘.’, skip: typing.SupportsInt = 0) -> None

Convert a CSV file to Parquet format.

Parameters

Name	Type	Description	Default
input_path	str	Path to the input CSV file.	required
output_path	str	Path to the output Parquet file.	required
compression	str	Compression codec: “zstd”, “snappy”, “lz4”, “gzip”, or “none”. Default is “zstd” if available, otherwise “gzip”.	required
row_group_size	int	Number of rows per row group. Default is 1,000,000.	required
num_threads	int	Number of threads to use. Default is auto-detect.	required
error_mode	str	Error handling mode: - “disabled” (default): No error collection, maximum performance - “fail_fast” or “strict”: Stop on first error - “permissive”: Collect all errors, stop on fatal - “best_effort”: Ignore errors, parse what’s possible	required
max_errors	int	Maximum number of errors to collect. Default is 10000. Setting this automatically enables “permissive” mode if error_mode is not set.	required
comment	str	String that marks comment lines. Lines starting with this string are skipped during parsing. Supports multi-character prefixes like “//” or “##”. Default is None (no comment skipping).	required
skip_empty_rows	bool	Whether to skip empty lines in the input. Default is True.	required
guess_integer	bool	Whether to infer integer types (INT32/INT64) for integer-like values. Default is True.	required
trim_ws	bool	Whether to trim leading and trailing whitespace from field values. Default is True.	required
escape_backslash	bool	Whether to use backslash escaping instead of doubled quotes. Default is False.	required
decimal_mark	str	Decimal separator character (‘.’ or ‘,’). Default is ‘.’.	required
skip	int	Number of lines to skip before the header row. Default is 0.	required

Raises

Name	Type	Description
	RuntimeError	If parsing fails. In permissive mode, collected errors are included in the exception message.

Examples

>>> import vroom_csv
>>> vroom_csv.to_parquet("data.csv", "data.parquet")

With error handling

>>> vroom_csv.to_parquet("data.csv", "data.parquet", error_mode="strict")