python关键字yield_python关键字yield的用法

用几个例子看看python的yield作用，顺带整理一下递归问题的求解。

Flatten问题：展开多层嵌套列表

对一个多层嵌套列表，flatten函数应该能按顺序打印出列表中的数字。

版本一：官方实现1

2

3

4

5

6

7def (nested):

try:

for sublist in nested:

for element in flatten2(sublist):

yield element

except TypeError:

yield nested

简洁精巧，但是将在下一个版本中展开递归的说明。

版本二：1

2

3

4

5

6

7

8from collections import Iterable

def flatten1(nested):

if not isinstance(nested, Iterable):

yield nested

else:

for sublist in nested:

for element in flatten1(sublist):

yield element

Target: get a flattened list from nested list

Method: using flatten(nested) to get a generator contains the flattened list of nested

Note: flatten(nested): generator # implies that result is type of generator

We can get recursion:flatten(node) = F( flatten(node.child_1), flatten(node.child_2), …, flatten(node.child_N) )

Then, what is F? We see the recursive case first:

LHS, flatten(node) should make a generator of the flattened node

RHS, node has many child(e.g. child_N). Assume that flatten(any_node) works and then,

we can get many generators(e.g. n-th generator is result of flatten(node.child_N)).

So, F should make a new generator that contains each element in these generators from child node. The core of recursion F can be code as:1

2

3for child_nested in nested: # traverse all child

for element in flatten(child_nested): # traverse element in a child generator

yield element

It is time to talk about base case (terminating case):

When the child_nested is a single element which is not iterable, it refers to the base case(leaf node). Now, for … in … clause will raise an exception because child_nested is not iterable.

flatten(leaf_node) should make a generator of only the single element. so, it can be coded as：1

2

3from collections import Iterable

if not isinstance(nested, Iterable):

yield nested

In this way, we finished coding the recursive function flatten().

My mistakes:omitted else: yield will continue to execute, so else can not be omitted.

base case should return a generator, so the recursive works. A generator makes another generator.

Eight queens puzzle

todo..

Summarypython中的关键字yield某种程度上相当于return(但是会从上次断开的地方继续执行)。

对递归不熟悉，总是陷于细节。

和数学中的函数类似，一个函数由定义域，值域和映射关系组成。通过整体去把握递归函数免于陷入细节。