import torch


def main():
    # print('===== Chapter 1 =====')
    # c1_introduction()
    # print('===== Chapter 2 =====')
    # c2_operator()
    # print('===== Chapter 3 =====')
    # c3_broadcast()
    # print('===== Chapter 4 =====')
    # c4_index_and_slice()
    # print('===== Chapter 5 =====')
    # c5_save_memory()
    print('===== Chapter 6 =====')
    c6_into_python_object()


def c1_introduction():
    x = torch.arange(12)
    print(f'x: {x}')
    print(f'x.shape: {x.shape}')
    print(f'x.numel(): {x.numel()}')
    xs = x.reshape(3, 4)
    print(f'x.reshape: {xs}')
    xs = x.reshape(-1, 4)
    print(f'x.reshape auto 1: {xs}')
    xs = x.reshape(3, -1)
    print(f'x.reshape auto 2: {xs}')

    zeros = torch.zeros((2, 3, 4))
    print(f'zeros: {zeros}')
    ones = torch.ones((2, 3, 4))
    print(f'ones: {ones}')

    randoms = torch.randn(3, 4)
    print(f'randn: {randoms}')

    manual = torch.tensor([[2, 1, 4, 3], [1, 2, 3, 4], [4, 3, 2, 1]], )
    print(f'manual: {manual}')
    # 看起来reshape第一个是行数
    manual = torch.tensor([2, 1, 4, 3, 1, 2, 3, 4, 4, 3, 2, 1]).reshape(3, -1)
    print(f'manual: {manual}')


def c2_operator():
    # torch按类型自动决定dtype
    x = torch.tensor([1.0, 2, 4, 8])
    print(x.dtype)
    x = torch.tensor([1, 2, 4, 8])
    print(x.dtype)

    # 强制指定dtype
    x = torch.tensor([1, 2, 4, 8], dtype=torch.float32)
    y = torch.tensor([2, 2, 2, 2])
    print(f'x + y: {x + y}')
    print(f'x - y: {x - y}')
    print(f'x * y: {x * y}')
    print(f'x / y: {x / y}')
    print(f'x ** y: {x ** y}')

    print(f'exp(x): {torch.exp(x)}')

    x = torch.arange(12, dtype=torch.float32).reshape(3, 4)
    y = torch.tensor([[2, 1, 4, 3], [1, 2, 3, 4], [4, 3, 2, 1]],
                     dtype=torch.float32)
    xy_row = torch.cat((x, y), dim=0)
    xy_col = torch.cat((x, y), dim=1)
    print(f'Row Cat: {xy_row}')
    print(f'Column Cat: {xy_col}')

    xy_equal = x == y
    print(f'Equal Boolean: {xy_equal}')

    x_sum = x.sum()
    print(f'x.sum: {x_sum}')


def c3_broadcast():
    a = torch.arange(3).reshape(3, -1)
    b = torch.arange(2).reshape(-1, 2)
    print(a)
    print(b)
    print(f'a + b: {a + b}')


def c4_index_and_slice():
    x = torch.arange(12, dtype=torch.float32).reshape(3, 4)
    print(x)
    print(x[-1])
    print(x[1:3])
    print(x[0::2])
    print(x[:, 0::2])

    x[1, 2] = 9
    print(x)
    x[:, 0::2] = 0
    print(x)
    y = torch.arange(6).reshape(-1, 2)
    x[:, 0::2] = y
    print(x)


def c5_save_memory():
    x = torch.arange(12, dtype=torch.float32).reshape(3, 4)
    y = torch.arange(12, dtype=torch.float32).reshape(3, 4)
    z = torch.zeros_like(x)
    z[:] = x + y
    print(z)

    z[:, :] = 0
    z[:] = x
    z += y
    print(z)


def c6_into_python_object():
    x = torch.arange(12, dtype=torch.float32).reshape(3, 4)
    a = x.numpy()
    print(type(a))
    b = torch.tensor(a)
    print(type(b))

    x = torch.tensor([3.5])
    print(x)
    print(x.item())
    print(float(x))
    print(int(x))


if __name__ == "__main__":
    main()