Function parse_file

python/markupever/parser.py:147–196

Parses an HTML or XML file and returns the parsed document tree. Args: path: A file path, file-like object, or Path object to be parsed. options: HTML or XML parsing options that control the parsing behavior. chunk_size: Size of chunks to read from the file during p

(
    path: typing.Union[str, typing.TextIO, typing.BinaryIO],
    options: typing.Union[
        _rustlib.HtmlOptions,
        _rustlib.XmlOptions,
        typing.Literal["html"],
        typing.Literal["xml"],
    ] = "html",
    *,
    chunk_size: int = 10240,
)

def parse_file(
    path: typing.Union[str, typing.TextIO, typing.BinaryIO],
    options: typing.Union[
        _rustlib.HtmlOptions,
        _rustlib.XmlOptions,
        typing.Literal["html"],
        typing.Literal["xml"],
    ] = "html",
    *,
    chunk_size: int = 10240,
) -> TreeDom:
    """
    Parses an HTML or XML file and returns the parsed document tree.

    Args:
        path: A file path, file-like object, or Path object to be parsed.
        options: HTML or XML parsing options that control the parsing behavior.
        chunk_size: Size of chunks to read from the file during parsing (default is 10240 bytes).

    Returns:
        A TreeDom object representing the parsed document tree.

    The function supports parsing files of different types (string paths, Path objects,
    file-like objects) and handles file opening and closing automatically.
    """
    from pathlib import Path

    close = False

    if isinstance(path, Path):
        path = str(path)

    if isinstance(path, str):
        path = open(path, "rb")
        close = True

    try:
        parser = Parser(options)

        while True:
            content = path.read(chunk_size)
            if not content:
                break

            parser.process(content)

        return parser.finish().into_dom()
    finally:
        if close:
            path.close()