Python 2 和 Python 3 列表推导的区别

Question

测试了一个列表推导，在 Python 3 中无法运行，大牛能否给个说明？

# python2  
matrix = [[1,2,3],[4,5,6],[7,8,9]]
[x for row in matrix for x in row]
# 值是 [1, 2, 3, 4, 5, 6, 7, 8, 9]
[x for x in row for row in matrix]
# 值是 [7, 7, 7, 8, 8, 8, 9, 9, 9]

# python3  
matrix = [[1,2,3],[4,5,6],[7,8,9]]
[x for row in matrix for x in row]
# 值是 [1, 2, 3, 4, 5, 6, 7, 8, 9]
[x for x in row for row in matrix]
# Traceback (most recent call last):
#  File "<stdin>", line 1, in <module>
# NameError: name 'row' is not defined

Answer 1

这和python没关系，你测试能运行，是因为你先运行了[x for row in matrix for x in row]，这时row变量是存在的所以没报错

Answer 2

你的第二种写法[x for x in row for row in matrix]本身就是错的，你在python2里先执行了[x for row in matrix for x in row]，这时候row就被声明了，不信你print(row)看看，或者你在python2里执行

matrix = [[1,2,3],[4,5,6],[7,8,9]]

[x for row in matrix for x in row]

[x for x in _row for _row in matrix]

看看最后一句报不报错