import os
import cv2
import mindspore.dataset as ds

class MySampler(ds.Sampler):
    def __iter__(self):
        if self.num_samples % 2 == 0:
            interval = 2
        elif self.num_samples % 3 == 0:
            interval = 3
        elif self.num_samples % 5 == 0:
            interval = 5
        else:
            interval = 1
        for i in range(0, self.num_samples, interval):
            yield i


class MyDataset:
    def __init__(self, dataset_dir):
        self.data_file = [os.path.join(dataset_dir, filename) for filename in os.listdir(dataset_dir)]

    def __getitem__(self, index):
        image = cv2.imread(self.data_file[index])
        return image

    def __len__(self):
        return len(self.data_file)

dataset_size = 20
dataset = ds.GeneratorDataset(MyDataset("./image"), column_names=["image"],
                              sampler=ds.IterSampler(MySampler(dataset_size)),
                              num_shards=8, shard_id=0)
for data in dataset.create_dict_iterator(output_numpy=True, num_epochs=1):
print(data)

dataset/test_s.py:31: in <module>
    dataset = ds.GeneratorDataset(MyDataset("./image"), column_names=["image"],
../../../mindspore/python/mindspore/dataset/engine/validators.py:1136: in new_method
    return method(self, *args, **kwargs)
../../../mindspore/python/mindspore/dataset/engine/datasets_user_defined.py:767: in __init__
    super().__init__(num_parallel_workers=num_parallel_workers, sampler=sampler, num_samples=num_samples,
../../../mindspore/python/mindspore/dataset/engine/datasets.py:2362: in __init__
    self.sampler = samplers.select_sampler(num_samples, sampler, shuffle, num_shards, shard_id)
../../../mindspore/python/mindspore/dataset/engine/samplers.py:55: in select_sampler
    raise ValueError(
E   ValueError: Conflicting arguments during sampler assignments. num_samples: None, num_shards: 8, shard_id: 0, shuffle: None.