jpayne@69: from __future__ import annotations
jpayne@69: 
jpayne@69: import email.utils
jpayne@69: import mimetypes
jpayne@69: import typing
jpayne@69: 
jpayne@69: _TYPE_FIELD_VALUE = typing.Union[str, bytes]
jpayne@69: _TYPE_FIELD_VALUE_TUPLE = typing.Union[
jpayne@69:     _TYPE_FIELD_VALUE,
jpayne@69:     typing.Tuple[str, _TYPE_FIELD_VALUE],
jpayne@69:     typing.Tuple[str, _TYPE_FIELD_VALUE, str],
jpayne@69: ]
jpayne@69: 
jpayne@69: 
jpayne@69: def guess_content_type(
jpayne@69:     filename: str | None, default: str = "application/octet-stream"
jpayne@69: ) -> str:
jpayne@69:     """
jpayne@69:     Guess the "Content-Type" of a file.
jpayne@69: 
jpayne@69:     :param filename:
jpayne@69:         The filename to guess the "Content-Type" of using :mod:`mimetypes`.
jpayne@69:     :param default:
jpayne@69:         If no "Content-Type" can be guessed, default to `default`.
jpayne@69:     """
jpayne@69:     if filename:
jpayne@69:         return mimetypes.guess_type(filename)[0] or default
jpayne@69:     return default
jpayne@69: 
jpayne@69: 
jpayne@69: def format_header_param_rfc2231(name: str, value: _TYPE_FIELD_VALUE) -> str:
jpayne@69:     """
jpayne@69:     Helper function to format and quote a single header parameter using the
jpayne@69:     strategy defined in RFC 2231.
jpayne@69: 
jpayne@69:     Particularly useful for header parameters which might contain
jpayne@69:     non-ASCII values, like file names. This follows
jpayne@69:     `RFC 2388 Section 4.4 <https://tools.ietf.org/html/rfc2388#section-4.4>`_.
jpayne@69: 
jpayne@69:     :param name:
jpayne@69:         The name of the parameter, a string expected to be ASCII only.
jpayne@69:     :param value:
jpayne@69:         The value of the parameter, provided as ``bytes`` or `str``.
jpayne@69:     :returns:
jpayne@69:         An RFC-2231-formatted unicode string.
jpayne@69: 
jpayne@69:     .. deprecated:: 2.0.0
jpayne@69:         Will be removed in urllib3 v2.1.0. This is not valid for
jpayne@69:         ``multipart/form-data`` header parameters.
jpayne@69:     """
jpayne@69:     import warnings
jpayne@69: 
jpayne@69:     warnings.warn(
jpayne@69:         "'format_header_param_rfc2231' is deprecated and will be "
jpayne@69:         "removed in urllib3 v2.1.0. This is not valid for "
jpayne@69:         "multipart/form-data header parameters.",
jpayne@69:         DeprecationWarning,
jpayne@69:         stacklevel=2,
jpayne@69:     )
jpayne@69: 
jpayne@69:     if isinstance(value, bytes):
jpayne@69:         value = value.decode("utf-8")
jpayne@69: 
jpayne@69:     if not any(ch in value for ch in '"\\\r\n'):
jpayne@69:         result = f'{name}="{value}"'
jpayne@69:         try:
jpayne@69:             result.encode("ascii")
jpayne@69:         except (UnicodeEncodeError, UnicodeDecodeError):
jpayne@69:             pass
jpayne@69:         else:
jpayne@69:             return result
jpayne@69: 
jpayne@69:     value = email.utils.encode_rfc2231(value, "utf-8")
jpayne@69:     value = f"{name}*={value}"
jpayne@69: 
jpayne@69:     return value
jpayne@69: 
jpayne@69: 
jpayne@69: def format_multipart_header_param(name: str, value: _TYPE_FIELD_VALUE) -> str:
jpayne@69:     """
jpayne@69:     Format and quote a single multipart header parameter.
jpayne@69: 
jpayne@69:     This follows the `WHATWG HTML Standard`_ as of 2021/06/10, matching
jpayne@69:     the behavior of current browser and curl versions. Values are
jpayne@69:     assumed to be UTF-8. The ``\\n``, ``\\r``, and ``"`` characters are
jpayne@69:     percent encoded.
jpayne@69: 
jpayne@69:     .. _WHATWG HTML Standard:
jpayne@69:         https://html.spec.whatwg.org/multipage/
jpayne@69:         form-control-infrastructure.html#multipart-form-data
jpayne@69: 
jpayne@69:     :param name:
jpayne@69:         The name of the parameter, an ASCII-only ``str``.
jpayne@69:     :param value:
jpayne@69:         The value of the parameter, a ``str`` or UTF-8 encoded
jpayne@69:         ``bytes``.
jpayne@69:     :returns:
jpayne@69:         A string ``name="value"`` with the escaped value.
jpayne@69: 
jpayne@69:     .. versionchanged:: 2.0.0
jpayne@69:         Matches the WHATWG HTML Standard as of 2021/06/10. Control
jpayne@69:         characters are no longer percent encoded.
jpayne@69: 
jpayne@69:     .. versionchanged:: 2.0.0
jpayne@69:         Renamed from ``format_header_param_html5`` and
jpayne@69:         ``format_header_param``. The old names will be removed in
jpayne@69:         urllib3 v2.1.0.
jpayne@69:     """
jpayne@69:     if isinstance(value, bytes):
jpayne@69:         value = value.decode("utf-8")
jpayne@69: 
jpayne@69:     # percent encode \n \r "
jpayne@69:     value = value.translate({10: "%0A", 13: "%0D", 34: "%22"})
jpayne@69:     return f'{name}="{value}"'
jpayne@69: 
jpayne@69: 
jpayne@69: def format_header_param_html5(name: str, value: _TYPE_FIELD_VALUE) -> str:
jpayne@69:     """
jpayne@69:     .. deprecated:: 2.0.0
jpayne@69:         Renamed to :func:`format_multipart_header_param`. Will be
jpayne@69:         removed in urllib3 v2.1.0.
jpayne@69:     """
jpayne@69:     import warnings
jpayne@69: 
jpayne@69:     warnings.warn(
jpayne@69:         "'format_header_param_html5' has been renamed to "
jpayne@69:         "'format_multipart_header_param'. The old name will be "
jpayne@69:         "removed in urllib3 v2.1.0.",
jpayne@69:         DeprecationWarning,
jpayne@69:         stacklevel=2,
jpayne@69:     )
jpayne@69:     return format_multipart_header_param(name, value)
jpayne@69: 
jpayne@69: 
jpayne@69: def format_header_param(name: str, value: _TYPE_FIELD_VALUE) -> str:
jpayne@69:     """
jpayne@69:     .. deprecated:: 2.0.0
jpayne@69:         Renamed to :func:`format_multipart_header_param`. Will be
jpayne@69:         removed in urllib3 v2.1.0.
jpayne@69:     """
jpayne@69:     import warnings
jpayne@69: 
jpayne@69:     warnings.warn(
jpayne@69:         "'format_header_param' has been renamed to "
jpayne@69:         "'format_multipart_header_param'. The old name will be "
jpayne@69:         "removed in urllib3 v2.1.0.",
jpayne@69:         DeprecationWarning,
jpayne@69:         stacklevel=2,
jpayne@69:     )
jpayne@69:     return format_multipart_header_param(name, value)
jpayne@69: 
jpayne@69: 
jpayne@69: class RequestField:
jpayne@69:     """
jpayne@69:     A data container for request body parameters.
jpayne@69: 
jpayne@69:     :param name:
jpayne@69:         The name of this request field. Must be unicode.
jpayne@69:     :param data:
jpayne@69:         The data/value body.
jpayne@69:     :param filename:
jpayne@69:         An optional filename of the request field. Must be unicode.
jpayne@69:     :param headers:
jpayne@69:         An optional dict-like object of headers to initially use for the field.
jpayne@69: 
jpayne@69:     .. versionchanged:: 2.0.0
jpayne@69:         The ``header_formatter`` parameter is deprecated and will
jpayne@69:         be removed in urllib3 v2.1.0.
jpayne@69:     """
jpayne@69: 
jpayne@69:     def __init__(
jpayne@69:         self,
jpayne@69:         name: str,
jpayne@69:         data: _TYPE_FIELD_VALUE,
jpayne@69:         filename: str | None = None,
jpayne@69:         headers: typing.Mapping[str, str] | None = None,
jpayne@69:         header_formatter: typing.Callable[[str, _TYPE_FIELD_VALUE], str] | None = None,
jpayne@69:     ):
jpayne@69:         self._name = name
jpayne@69:         self._filename = filename
jpayne@69:         self.data = data
jpayne@69:         self.headers: dict[str, str | None] = {}
jpayne@69:         if headers:
jpayne@69:             self.headers = dict(headers)
jpayne@69: 
jpayne@69:         if header_formatter is not None:
jpayne@69:             import warnings
jpayne@69: 
jpayne@69:             warnings.warn(
jpayne@69:                 "The 'header_formatter' parameter is deprecated and "
jpayne@69:                 "will be removed in urllib3 v2.1.0.",
jpayne@69:                 DeprecationWarning,
jpayne@69:                 stacklevel=2,
jpayne@69:             )
jpayne@69:             self.header_formatter = header_formatter
jpayne@69:         else:
jpayne@69:             self.header_formatter = format_multipart_header_param
jpayne@69: 
jpayne@69:     @classmethod
jpayne@69:     def from_tuples(
jpayne@69:         cls,
jpayne@69:         fieldname: str,
jpayne@69:         value: _TYPE_FIELD_VALUE_TUPLE,
jpayne@69:         header_formatter: typing.Callable[[str, _TYPE_FIELD_VALUE], str] | None = None,
jpayne@69:     ) -> RequestField:
jpayne@69:         """
jpayne@69:         A :class:`~urllib3.fields.RequestField` factory from old-style tuple parameters.
jpayne@69: 
jpayne@69:         Supports constructing :class:`~urllib3.fields.RequestField` from
jpayne@69:         parameter of key/value strings AND key/filetuple. A filetuple is a
jpayne@69:         (filename, data, MIME type) tuple where the MIME type is optional.
jpayne@69:         For example::
jpayne@69: 
jpayne@69:             'foo': 'bar',
jpayne@69:             'fakefile': ('foofile.txt', 'contents of foofile'),
jpayne@69:             'realfile': ('barfile.txt', open('realfile').read()),
jpayne@69:             'typedfile': ('bazfile.bin', open('bazfile').read(), 'image/jpeg'),
jpayne@69:             'nonamefile': 'contents of nonamefile field',
jpayne@69: 
jpayne@69:         Field names and filenames must be unicode.
jpayne@69:         """
jpayne@69:         filename: str | None
jpayne@69:         content_type: str | None
jpayne@69:         data: _TYPE_FIELD_VALUE
jpayne@69: 
jpayne@69:         if isinstance(value, tuple):
jpayne@69:             if len(value) == 3:
jpayne@69:                 filename, data, content_type = value
jpayne@69:             else:
jpayne@69:                 filename, data = value
jpayne@69:                 content_type = guess_content_type(filename)
jpayne@69:         else:
jpayne@69:             filename = None
jpayne@69:             content_type = None
jpayne@69:             data = value
jpayne@69: 
jpayne@69:         request_param = cls(
jpayne@69:             fieldname, data, filename=filename, header_formatter=header_formatter
jpayne@69:         )
jpayne@69:         request_param.make_multipart(content_type=content_type)
jpayne@69: 
jpayne@69:         return request_param
jpayne@69: 
jpayne@69:     def _render_part(self, name: str, value: _TYPE_FIELD_VALUE) -> str:
jpayne@69:         """
jpayne@69:         Override this method to change how each multipart header
jpayne@69:         parameter is formatted. By default, this calls
jpayne@69:         :func:`format_multipart_header_param`.
jpayne@69: 
jpayne@69:         :param name:
jpayne@69:             The name of the parameter, an ASCII-only ``str``.
jpayne@69:         :param value:
jpayne@69:             The value of the parameter, a ``str`` or UTF-8 encoded
jpayne@69:             ``bytes``.
jpayne@69: 
jpayne@69:         :meta public:
jpayne@69:         """
jpayne@69:         return self.header_formatter(name, value)
jpayne@69: 
jpayne@69:     def _render_parts(
jpayne@69:         self,
jpayne@69:         header_parts: (
jpayne@69:             dict[str, _TYPE_FIELD_VALUE | None]
jpayne@69:             | typing.Sequence[tuple[str, _TYPE_FIELD_VALUE | None]]
jpayne@69:         ),
jpayne@69:     ) -> str:
jpayne@69:         """
jpayne@69:         Helper function to format and quote a single header.
jpayne@69: 
jpayne@69:         Useful for single headers that are composed of multiple items. E.g.,
jpayne@69:         'Content-Disposition' fields.
jpayne@69: 
jpayne@69:         :param header_parts:
jpayne@69:             A sequence of (k, v) tuples or a :class:`dict` of (k, v) to format
jpayne@69:             as `k1="v1"; k2="v2"; ...`.
jpayne@69:         """
jpayne@69:         iterable: typing.Iterable[tuple[str, _TYPE_FIELD_VALUE | None]]
jpayne@69: 
jpayne@69:         parts = []
jpayne@69:         if isinstance(header_parts, dict):
jpayne@69:             iterable = header_parts.items()
jpayne@69:         else:
jpayne@69:             iterable = header_parts
jpayne@69: 
jpayne@69:         for name, value in iterable:
jpayne@69:             if value is not None:
jpayne@69:                 parts.append(self._render_part(name, value))
jpayne@69: 
jpayne@69:         return "; ".join(parts)
jpayne@69: 
jpayne@69:     def render_headers(self) -> str:
jpayne@69:         """
jpayne@69:         Renders the headers for this request field.
jpayne@69:         """
jpayne@69:         lines = []
jpayne@69: 
jpayne@69:         sort_keys = ["Content-Disposition", "Content-Type", "Content-Location"]
jpayne@69:         for sort_key in sort_keys:
jpayne@69:             if self.headers.get(sort_key, False):
jpayne@69:                 lines.append(f"{sort_key}: {self.headers[sort_key]}")
jpayne@69: 
jpayne@69:         for header_name, header_value in self.headers.items():
jpayne@69:             if header_name not in sort_keys:
jpayne@69:                 if header_value:
jpayne@69:                     lines.append(f"{header_name}: {header_value}")
jpayne@69: 
jpayne@69:         lines.append("\r\n")
jpayne@69:         return "\r\n".join(lines)
jpayne@69: 
jpayne@69:     def make_multipart(
jpayne@69:         self,
jpayne@69:         content_disposition: str | None = None,
jpayne@69:         content_type: str | None = None,
jpayne@69:         content_location: str | None = None,
jpayne@69:     ) -> None:
jpayne@69:         """
jpayne@69:         Makes this request field into a multipart request field.
jpayne@69: 
jpayne@69:         This method overrides "Content-Disposition", "Content-Type" and
jpayne@69:         "Content-Location" headers to the request parameter.
jpayne@69: 
jpayne@69:         :param content_disposition:
jpayne@69:             The 'Content-Disposition' of the request body. Defaults to 'form-data'
jpayne@69:         :param content_type:
jpayne@69:             The 'Content-Type' of the request body.
jpayne@69:         :param content_location:
jpayne@69:             The 'Content-Location' of the request body.
jpayne@69: 
jpayne@69:         """
jpayne@69:         content_disposition = (content_disposition or "form-data") + "; ".join(
jpayne@69:             [
jpayne@69:                 "",
jpayne@69:                 self._render_parts(
jpayne@69:                     (("name", self._name), ("filename", self._filename))
jpayne@69:                 ),
jpayne@69:             ]
jpayne@69:         )
jpayne@69: 
jpayne@69:         self.headers["Content-Disposition"] = content_disposition
jpayne@69:         self.headers["Content-Type"] = content_type
jpayne@69:         self.headers["Content-Location"] = content_location