Class MapData

tensorpack/dataflow/common.py:287–318

Apply a mapper/filter on the datapoints of a DataFlow. Note: 1. Please make sure func doesn't modify its arguments in place, unless you're certain it's safe. 2. If you discard some datapoints, ``len(MapData(ds))`` will be incorrect. Example: .. code

class MapData(ProxyDataFlow):
    """
    Apply a mapper/filter on the datapoints of a DataFlow.

    Note:
        1. Please make sure func doesn't modify its arguments in place,
           unless you're certain it's safe.
        2. If you discard some datapoints, ``len(MapData(ds))`` will be incorrect.

    Example:

        .. code-block:: none

            ds = Mnist('train')  # each datapoint is [img, label]
            ds = MapData(ds, lambda dp: [dp[0] * 255, dp[1]])
    """

    def __init__(self, ds, func):
        """
        Args:
            ds (DataFlow): input DataFlow
            func (datapoint -> datapoint | None): takes a datapoint and returns a new
                datapoint. Return None to discard/skip this datapoint.
        """
        super(MapData, self).__init__(ds)
        self.func = func

    def __iter__(self):
        for dp in self.ds:
            ret = self.func(copy(dp))  # shallow copy the list
            if ret is not None:
                yield ret


class MapDataComponent(MapData):