创建 X 数据的解析工具

new file:   utils/__init__.py
new file:   utils/dataset.py 通用数据工具
new file:   utils/xdecode.py 解析 X.h5 数据

utils/dataset.py

@@ -0,0 +1,20 @@
+import pickle
+
+
+class Bunch(dict):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.__dict__ = self
+
+
+def bunch2json(bunch, path):
+    # bunch 序列化为 JSON
+    with open(path, 'wb') as fp:
+        pickle.dump(bunch, fp)
+
+
+def json2bunch(path):
+    # JSON 反序列化为 bunch
+    with open(path, 'rb') as fp:
+        X = pickle.load(fp)
+    return X

utils/xdecode.py

@@ -0,0 +1,111 @@
+import numpy as np
+import tables as tb
+
+
+class XDecode:
+    def __init__(self, root='../datasome/X.h5'):
+        '''解析数据 X
+        使用完该实例记得关闭 self.h5.close()
+        
+        实例属性
+        ========
+        members 即 ['mnist', 'fashion_mnist', 'cifar100', 'cifar10']
+        '''
+        self.h5 = tb.open_file(root)
+        self.members = self.h5.root.__members__
+
+    def summary(self):
+        '''打印 X 数据的摘要信息'''
+        print(self.h5)
+
+    def get_members(self, name):
+        '''获得给定 name 的成员名称列表
+        
+        参数
+        =======
+        name in ['mnist', 'fashion_mnist', 'cifar100', 'cifar10']
+        '''
+        return self.h5.root[name].__members__
+
+    def get_mnist_members(self):
+        return self.h5.root.mnist.__members__
+
+    def get_trainX(self, name):
+        '''获得给定 name 的 trainX
+        
+        参数
+        =======
+        name in ['mnist', 'fashion_mnist', 'cifar100', 'cifar10']
+        '''
+        return self.h5.get_node(f'/{name}', 'trainX')
+
+    def get_testX(self, name):
+        '''获得给定 name 的 testX
+        
+        参数
+        =======
+        name in ['mnist', 'fashion_mnist', 'cifar100', 'cifar10']
+        '''
+        return self.h5.get_node(f'/{name}', 'testX')
+
+    def get_trainY(self, name):
+        '''获得给定 name 的 trainY
+        
+        参数
+        =======
+        name in ['mnist', 'fashion_mnist', 'cifar10']
+        '''
+        return self.h5.get_node(f'/{name}', 'trainY')
+
+    def get_testY(self, name):
+        '''获得给定 name 的 testY
+        
+        参数
+        =======
+        name in ['mnist', 'fashion_mnist', 'cifar10']
+        '''
+        return self.h5.get_node(f'/{name}', 'testY')
+
+    def get_train_coarse_labels(self):
+        '''获得 Cifar100 训练集的粗标签'''
+        return self.h5.get_node('/cifar100', 'train_coarse_labels')
+
+    def get_train_fine_labels(self):
+        '''获得 Cifar100 训练集的细标签'''
+        return self.h5.get_node('/cifar100', 'train_fine_labels')
+
+    def get_test_coarse_labels(self):
+        '''获得 Cifar100 测试集的粗标签'''
+        return self.h5.get_node('/cifar100', 'test_coarse_labels')
+
+    def get_test_fine_labels(self):
+        '''获得 Cifar100 的测试集细标签'''
+        return self.h5.get_node('/cifar100', 'test_fine_labels')
+
+    def get_coarse_label_names(self):
+        '''获得 Cifar100 测试集的粗标签的名称'''
+        label_names = self.h5.get_node('/cifar100', 'coarse_label_names')
+        return np.asanyarray(label_names, "U")
+
+    def get_fine_label_names(self):
+        '''获得 Cifar100 的测试集细标签的名称'''
+        label_names = self.h5.get_node('/cifar100', 'fine_label_names')
+        return np.asanyarray(label_names, "U")
+
+    def get_label_names(self, name):
+        '''获得给定 name 的标签名称
+        
+        参数
+        ======
+        name in ['mnist', 'fashion_mnist', 'cifar10']
+        '''
+        if name == 'cifar10':
+            label_names = self.h5.get_node('/cifar10', 'label_names')
+            return np.asanyarray(label_names, "U")
+        elif name == 'fashion_mnist':
+            return [
+                'T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
+                'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot'
+            ]
+        elif name == 'mnist':
+            return np.arange(10)