Method as_vector

bson/binary.py:544–633

From the Binary, create a list or 1-d numpy array of numbers, along with dtype and padding. :param return_numpy: If True, BinaryVector.data will be a one-dimensional numpy array. By default, it is a list. :return: BinaryVector .. versionadded:: 4.10

(self, return_numpy: bool = False)

        return cls(metadata + data, subtype=VECTOR_SUBTYPE)

    def as_vector(self, return_numpy: bool = False) -> BinaryVector:
        """From the Binary, create a list or 1-d numpy array of numbers, along with dtype and padding.

        :param return_numpy: If True, BinaryVector.data will be a one-dimensional numpy array. By default, it is a list.
        :return: BinaryVector

        .. versionadded:: 4.10
        """

        if self.subtype != VECTOR_SUBTYPE:
            raise ValueError(f"Cannot decode subtype {self.subtype} as a vector")

        dtype, padding = struct.unpack_from("<sB", self)
        dtype = BinaryVectorDtype(dtype)
        offset = 2
        n_bytes = len(self) - offset

        if padding and dtype != BinaryVectorDtype.PACKED_BIT:
            raise ValueError(
                f"Corrupt data. Padding ({padding}) must be 0 for all but PACKED_BIT dtypes. ({dtype=})"
            )

        if not return_numpy:
            if dtype == BinaryVectorDtype.INT8:
                dtype_format = "b"
                format_string = f"<{n_bytes}{dtype_format}"
                vector = list(struct.unpack_from(format_string, self, offset))
                return BinaryVector(vector, dtype, padding)

            elif dtype == BinaryVectorDtype.FLOAT32:
                n_values = n_bytes // 4
                if n_bytes % 4:
                    raise ValueError(
                        "Corrupt data. N bytes for a float32 vector must be a multiple of 4."
                    )
                dtype_format = "f"
                format_string = f"<{n_values}{dtype_format}"
                vector = list(struct.unpack_from(format_string, self, offset))
                return BinaryVector(vector, dtype, padding)

            elif dtype == BinaryVectorDtype.PACKED_BIT:
                # data packed as uint8
                if padding and not n_bytes:
                    raise ValueError("Corrupt data. Vector has a padding P, but no data.")
                if padding > 7 or padding < 0:
                    raise ValueError(f"Corrupt data. Padding ({padding}) must be between 0 and 7.")
                dtype_format = "B"
                format_string = f"<{n_bytes}{dtype_format}"
                unpacked_uint8s = list(struct.unpack_from(format_string, self, offset))
                if padding and n_bytes and unpacked_uint8s[-1] & (1 << padding) - 1 != 0:
                    warnings.warn(
                        "Vector has a padding P, but bits in the final byte lower than P are non-zero. For pymongo>=5.0, they must be zero.",
                        DeprecationWarning,
                        stacklevel=2,
                    )
                return BinaryVector(unpacked_uint8s, dtype, padding)

            else: