Python中多层List展平为一层

小书匠python

使用Python脚本的过程中,偶尔需要使用list多层转一层,又总是忘记怎么写搜索关键词,所以总是找了很久,现在把各种方法记录下来,方便自己也方便大家.

方法很多,现在就简单写8种,后面再对这8种方法做基准测试.

声明:文中的方法均收集自Making a flat list out of list of lists in Python

---

1.定义减层方法

1. import functools 
   

   ---

2. import itertools 
   

   ---

3. import numpy 
   

   ---

4. import operator 
   

   ---

5. import perfplot 
   

   ---

6. from collections import Iterable # or from collections.abc import Iterable 
   

   ---

7. from iteration_utilities import deepflatten 
   

   ---

8. ---

9. #使用两次for循环 
   

   ---

10. def forfor(a): 
    

    ---

11. return [item for sublist in a for item in sublist] 
    

    ---

12. ---

13. #通过sum 
    

    ---

14. def sum_brackets(a): 
    

    ---

15. return sum(a, []) 
    

    ---

16. ---

17. #使用functools內建模块 
    

    ---

18. def functools_reduce(a): 
    

    ---

19. return functools.reduce(operator.concat, a) 
    

    ---

20. ---

21. #使用itertools內建模块 
    

    ---

22. def itertools_chain(a): 
    

    ---

23. return list(itertools.chain.from_iterable(a)) 
    

    ---

24. ---

25. #使用numpy 
    

    ---

26. def numpy_flat(a): 
    

    ---

27. return list(numpy.array(a).flat) 
    

    ---

28. ---

29. #使用numpy 
    

    ---

30. def numpy_concatenate(a): 
    

    ---

31. return list(numpy.concatenate(a)) 
    

    ---

32. ---

33. #自定义函数 
    

    ---

34. def flatten(items): 
    

    ---

35. """Yield items from any nested iterable; see REF.""" 
    

    ---

36. for x in items: 
    

    ---

37. if isinstance(x, Iterable) and not isinstance(x, (str, bytes)): 
    

    ---

38. yield from flatten(x) 
    

    ---

39. else: 
    

    ---

40. yield x 
    

    ---

41. ---

42. def pylangs_flatten(a): 
    

    ---

43. return list(flatten(a)) 
    

    ---

44. ---

45. #使用库iteration_utilities 
    

    ---

46. def iteration_utilities_deepflatten(a): 
    

    ---

47. return list(deepflatten(a, depth=1)) 
    

    ---

2.测试

1. a=[[1,2,3],[4,5,6],[7,8,9]] 
   

   ---

2. print(a) 
   

   ---

3. ---

4. print('--------------------------') 
   

   ---

5. ---

6. print(forfor(a)) 
   

   ---

7. print(sum_brackets(a)) 
   

   ---

8. print(functools_reduce(a)) 
   

   ---

9. print(itertools_chain(a)) 
   

   ---

10. print(numpy_flat(a)) 
    

    ---

11. print(numpy_concatenate(a)) 
    

    ---

12. print(pylangs_flatten(a)) 
    

    ---

13. print(iteration_utilities_deepflatten(a)) 
    

    ---

输出:

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

--------------------------

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

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

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

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

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

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

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

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

---

2.各种方法的基准测试(消耗时间对比)

各种方法在小数据上消耗时间差别不大,如果数据很小,没必要为了选择而烦恼,如果数据很大,可以参考下面基准测试的结果来选择减层方法.

1. import matplotlib.pyplot as plt 
   

   ---

2. from simple_benchmark import benchmark 
   

   ---

3. ---

4. #基准测试 
   

   ---

5. b = benchmark( 
   

   ---

6. [forfor, sum_brackets, functools_reduce, itertools_chain,numpy_flat, numpy_concatenate, pylangs_flatten,iteration_utilities_deepflatten], 
   

   ---

7. arguments={2**i: [[0]*5]*(2**i) for i in range(1, 13)}, 
   

   ---

8. argument_name='number of inner lists' 
   

   ---

9. ) 
   

   ---

10. ---

11. #显示测试结果 
    

    ---

12. plt.subplots(1,1,figsize=(15,10)) 
    

    ---

13. b.plot() 
    

    ---

14. plt.legend(loc = 'upper left') 
    

    ---

15. plt.show() 
    

    ---

消耗时间对比

相同数据量,纵轴方向越小,方法越快.

---

代码可以从这里下载,需要部署Jupyter环境,可参考我的博客部署方法.