如何找到字符串中任何一组字符的第一个索引

How to find the first index of any of a set of characters in a string

我想找到字符串中任何“特殊”字符第一次出现的索引,如下所示:

>>> "Hello world!".index([' ', '!'])
5

…除了那不是有效的 Python 语法。当然,我可以编写一个函数来模拟这种行为:

def first_index(s, characters):
    i = []
    for c in characters:
        try:
            i.append(s.index(c))
        except ValueError:
            pass
    if not i:
        raise ValueError
    return min(i)

我也可以使用正则表达式,但这两种解决方案似乎都有些矫枉过正。在 Python 中是否有任何“理智”的方法来做到这一点?

使用 gen-exp 和 find 方法。

>>> a = [' ', '!']
>>> s = "Hello World!"
>>> min(s.find(i) for i in a)
5

若要删除 -1,您可以在列表 comp

中设置过滤器
>>> a = [' ', '!','$']
>>> s = "Hello World!"
>>> min(s.find(i) for i in a if i in s)
5

或者你可以替换 None

>>> min(s.find(i) if i in s else None for i in a)
5

添加 timeit 个结果

$ python -m timeit "a = [' ', '\!'];s = 'Hello World\!';min(s.find(i) for i in a if i in s)"
1000000 loops, best of 3: 0.902 usec per loop
$ python -m timeit "a = [' ', '\!'];s = 'Hello World\!';next((i for i, ch  in enumerate(s) if ch in a),None)"
1000000 loops, best of 3: 1.25 usec per loop
$ python -m timeit "a = [' ', '\!'];s = 'Hello World\!';min(map(lambda x: (s.index(x) if (x in s) else len(s)), a))"
1000000 loops, best of 3: 1.12 usec per loop

在您的 Example 案例中,Padraic 的漂亮解决方案有点慢。但是在大型测试用例中,它绝对是赢家。 (有点意外的是alfasin的"Not as optimized"这里也更快)

添加了实施细节

>>> def take1(s,a):
...     min(s.find(i) for i in a if i in s)
... 
>>> import dis
>>> dis.dis(take1)
  2           0 LOAD_GLOBAL              0 (min)
              3 LOAD_CLOSURE             0 (s)
              6 BUILD_TUPLE              1
              9 LOAD_CONST               1 (<code object <genexpr> at 0x7fa622e961b0, file "<stdin>", line 2>)
             12 MAKE_CLOSURE             0
             15 LOAD_FAST                1 (a)
             18 GET_ITER            
             19 CALL_FUNCTION            1
             22 CALL_FUNCTION            1
             25 POP_TOP             
             26 LOAD_CONST               0 (None)
             29 RETURN_VALUE        
>>> def take2(s,a):
...     next((i for i, ch  in enumerate(s) if ch in a),None)
... 
>>> dis.dis(take2)
  2           0 LOAD_GLOBAL              0 (next)
              3 LOAD_CLOSURE             0 (a)
              6 BUILD_TUPLE              1
              9 LOAD_CONST               1 (<code object <genexpr> at 0x7fa622e96e30, file "<stdin>", line 2>)
             12 MAKE_CLOSURE             0
             15 LOAD_GLOBAL              1 (enumerate)
             18 LOAD_FAST                0 (s)
             21 CALL_FUNCTION            1
             24 GET_ITER            
             25 CALL_FUNCTION            1
             28 LOAD_CONST               0 (None)
             31 CALL_FUNCTION            2
             34 POP_TOP             
             35 LOAD_CONST               0 (None)
             38 RETURN_VALUE        
>>> def take3(s,a):
...     min(map(lambda x: (s.index(x) if (x in s) else len(s)), a))
... 
>>> dis.dis(take3)
  2           0 LOAD_GLOBAL              0 (min)
              3 LOAD_GLOBAL              1 (map)
              6 LOAD_CLOSURE             0 (s)
              9 BUILD_TUPLE              1
             12 LOAD_CONST               1 (<code object <lambda> at 0x7fa622e44eb0, file "<stdin>", line 2>)
             15 MAKE_CLOSURE             0
             18 LOAD_FAST                1 (a)
             21 CALL_FUNCTION            2
             24 CALL_FUNCTION            1
             27 POP_TOP             
             28 LOAD_CONST               0 (None)
             31 RETURN_VALUE        

正如您在 Padraic 的案例中清楚地看到的那样,全局函数 nextenumerate 的加载与最后的 None 一起消磨时间。在 alfasin 的解决方案中,主要的减速是 lambda 函数。

您可以使用 enumerate and next with a generator expression,获取第一个匹配项或 returning None 如果没有字符出现在 s:

s = "Hello world!"

st = {"!"," "}
ind = next((i for i, ch  in enumerate(s) if ch in st),None)
print(ind)

如果没有匹配项,您可以将您想要的任何值作为默认值 return 传递给下一个值。

如果您想使用函数并引发 ValueError:

def first_index(s, characters):
    st = set(characters)
    ind = next((i for i, ch in enumerate(s) if ch in st), None)
    if ind is not None:
        return ind
    raise ValueError

对于较小的输入,使用集合不会有太大区别,但对于较大的字符串,它会更有效。

一些时间:

字符串中,字符集的最后一个字符:

In [40]: s = "Hello world!" * 100    
In [41]: string = s    
In [42]: %%timeit
st = {"x","y","!"}
next((i for i, ch in enumerate(s) if ch in st), None)
   ....: 
1000000 loops, best of 3: 1.71 µs per loop    
In [43]: %%timeit
specials = ['x', 'y', '!']
min(map(lambda x: (string.index(x) if (x in string) else len(string)), specials))
   ....: 
100000 loops, best of 3: 2.64 µs per loop

不在字符串中,更大的字符集:

In [44]: %%timeit
st = {"u","v","w","x","y","z"}
next((i for i, ch in enumerate(s) if ch in st), None)
   ....: 
1000000 loops, best of 3: 1.49 µs per loop

In [45]: %%timeit
specials = ["u","v","w","x","y","z"]
min(map(lambda x: (string.index(x) if (x in string) else len(string)), specials))
   ....: 
100000 loops, best of 3: 5.48 µs per loop

字符串中字符集的第一个字符:

In [47]: %%timeit
specials = ['H', 'y', '!']
min(map(lambda x: (string.index(x) if (x in string) else len(string)), specials))
   ....: 
100000 loops, best of 3: 2.02 µs per loop

In [48]: %%timeit
st = {"H","y","!"}
next((i for i, ch in enumerate(s) if ch in st), None)
   ....: 
1000000 loops, best of 3: 903 ns per loop

不像 Padraic Cunningham 的解决方案那么优化,但仍然是一个线性:

string = "Hello world!"
specials = [' ', '!', 'x']
min(map(lambda x: (string.index(x) if (x in string) else len(string)), specials))

我更喜欢 re 模块,因为它是内置的并且已经过测试。它也针对这类事情进行了优化。

>>> import re
>>> re.search(r'[ !]', 'Hello World!').start()
5

您可能想要检查是否找到了匹配项,或者在未找到匹配项时捕获异常。

不使用 re 是有原因的,但我希望看到一个很好的评论来证明合理性。认为自己可以 "do it better" 通常是不必要的,这会使其他人更难阅读代码并且更难维护。