转自 http://www.cnblogs.com/BeginMan/p/3178103.html
一、关于函数式编程的内建函数
apply()逐渐被舍弃，这里不讨论
1、filter()
#filter(func,seq)
"""纯Python描述filter函数"""
def Myfilter(bool_func,seq): filtered_seq = [] for obj in seq: if bool_func(obj): filtered_seq.append(obj) return filtered_seq print Myfilter(lambda x:x%3==0, [1,2,3,4,5,6,7,8,9]) #[3, 6, 9] """Build-In function filter()""" print filter(lambda x:x%3==0, [1,2,3,4,5,6,7,8,9]) #[3, 6, 9] """非函数下的实现""" print [x for x in [1,2,3,4,5,6,7,8,9] if x%3==0] #[3, 6, 9] #还记得上一节写的，如果能有for..in..if（或列表解析）能处理的最好用这个而放弃lambda
试验：列表解析和lambda性能比较
lambda:
import time
start = time.clock()
filter(lambda x:x%3==0, [i for i in range(10000000)]) end = time.clock() print end-start #耗时4.9441799282
列表解析：
import time
start = time.clock()
[x for x in range(10000000) if x%3==0] end = time.clock() print end-start #耗时2.95589058109
从上可见，最好使用列表解析。
2、map()
"""纯Python模拟map()"""
def Mymap(func,seq):
mapped_seq = [] for obj in seq: mapped_seq.append(func(obj)) return mapped_seq print Mymap(lambda x:x*10,[i for i in range(10)]) #[0, 10, 20, 30, 40, 50, 60, 70, 80, 90] """内建map()函数""" print map(lambda x:x*10,[i for i in range(10)]) #[0, 10, 20, 30, 40, 50, 60, 70, 80, 90] """非函数式编程""" print [x*10 for x in range(10)] #[0, 10, 20, 30, 40, 50, 60, 70, 80, 90] """多个序列的map()""" print map(lambda x,y:x+y,[1,2,3],[4,5,6]) #[5, 7, 9] """None与map()""" print map(None,[1,2,3],[4,5,6]) #[(1, 4), (2, 5), (3, 6)] """None在map()中使用效果同zip()，将不相干的序列归并在一起""" print zip([1,2,3],[4,5,6]) #[(1, 4), (2, 5), (3, 6)]
3、reduce()
"""纯Python模拟reduce()"""
def Myreduce(bin_func,seq,init=None):
Iseq = list(seq) if init is None: res = Iseq.pop() else: res = init for obj in Iseq: res = bin_func(res,obj) return res print Myreduce(lambda x,y:x+y, [i for i in range(10)]) #45 print Myreduce(lambda x,y:x+y, [i for i in range(10)],100) #145 """reduce()""" print reduce(lambda x,y:x+y,[i for i in range(10)]) #45 print reduce(lambda x,y:x+y,[i for i in range(10)],100) #145
综上，尽量用最简便的方式去实现。