STL可能的误用-find_first_of和erase

一.string中find_first_of的误用
STL中提供的string可以说极大方便了对字符串的操作,但是很多函数由于样子上很相似,所以导致很容易理解错误,find_first_of和find就是一个很好的例子。
我们先来看一下string提供的查找相关的函数列表:

1
2
3
4
5
find_first_of() 查找第一个与value中的某值相等的字符
find_first_not_of() 查找第一个与value中的所有值都不相等的字符
find_last_of() 查找最后一个与value中的某值相等的字符
find_last_not_of() 查找最后一个与value中的所有值都不相等的字符
rfind() 查找最后一个与value相等的字符(逆向查找)

如此简洁的说明,其实完全没有把他们最重要的区别描述出来,请务必记住:
对于find和rfind:

  • 匹配的是整个被查找串

对于find_first_of,find_first_not_of,find_last_of,find_last_not_of:

  • 匹配的是被查找串中的任意字符

我们来测试一下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
#include <iostream>
#include <memory>
#include <string>
#include <vector>
#include <set>
#include <map>
using namespace std;
int main(int argc, const char *argv[])
{
    string src = "vimer.cn";
 
    string str1 = "mer";
    string str2 = "sre";
 
    size_t pos;
 
    pos = src.find(str1);
    cout<<pos<<endl;
 
    pos = src.find(str2);
    cout<<pos<<endl;
 
    pos = src.find_first_of(str1);
    cout<<pos<<endl;
 
    pos = src.find_first_of(str2);
    cout<<pos<<endl;
 
    return 0;
}

运行结果如下:

2
4294967295
2
3

结果中4294967295即string::npos,代表没有找到。而在find_first_of(str2)的时候返回的是3,即字符'e',证明了我们之前的说法。

二.erase函数的误用
STL的容器一般都会提供erase方法,vector,list等的erase方法都会返回下一个元素的指针,所以在for循环中对容器进行erase时,代码一般如下即可:

1
2
3
4
5
6
7
8
9
10
11
for(vector<uint32_t>::iterator it = myVec.begin(); it != myVec.end();)
{
    if (it->data == data)
    {
        it = myVec.erase(it);
    }
    else
    {
        ++it;
    }
}

但是偏偏stl的map的erase不会返回下一个的指针,所以使用起来很让人困惑。其实本来很简单的一个问题,现在却被搞得很复杂。
为了不给大家造成混淆,这里只列出两种正确的方法,大家按照这种方法来写就绝对没有问题,也不用考虑不同的stl的map实现。
1.简短型

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
#include <iostream>
#include <memory>
#include <string>
#include <vector>
#include <set>
#include <map>
using namespace std;
int main(int argc, const char *argv[])
{
    map<unsigned int,int> myMap;
    int count = 10;
    for (int i = 0; i < count; i++)
    {
        myMap[i]=i;
    }
    for(map<unsigned int, int>::iterator it = myMap.begin(); it != myMap.end(); )
    {
        if (it->first == 3 || it->first == 9)
        {
            myMap.erase(it++);
        }
        else
        {
            it++;
        }
    }
    for(map<unsigned int, int>::iterator it = myMap.begin(); it != myMap.end(); ++it)
    {
        cout<<it->second<<endl;
    }
    return 0;
}

2.易读型

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
#include <iostream>
#include <memory>
#include <string>
#include <vector>
#include <set>
#include <map>
using namespace std;
int main(int argc, const char *argv[])
{
    map<unsigned int,int> myMap;
    int count = 10;
    for (int i = 0; i < count; i++)
    {
        myMap[i]=i;
    }
    for(map<unsigned int, int>::iterator it = myMap.begin(); it != myMap.end(); )
    {
        map<unsigned int, int>::iterator tempit = it;
        it++;
 
        if (tempit->first == 3 || tempit->first == 9)
        {
            myMap.erase(tempit);
        }
    }
    for(map<unsigned int, int>::iterator it = myMap.begin(); it != myMap.end(); ++it)
    {
        cout<<it->second<<endl;
    }
    return 0;
}

运行结果都为:

0
1
2
4
5
6
7
8

对于第一种方法,千万不要理解等同于:

1
2
3
4
5
6
//这样是错误的,不要模仿!
if (it->first == 3 || it->first == 9)
{
    myMap.erase(it);
}
it++;

对于笔者本人来说,更倾向第二种方法,因为虽然文章中是直接调用

1
myMap.erase(tempit);

但实际情况可能是调用一个函数,而在这个函数里面会有一堆逻辑出来判断是否要删除这个元素,这种情况下只有用第二种方法能够满足。

OK,STL的强大和危险性是成正比的,所以要熟练运用还是要深入理解才行,希望能和大家一起进步~





原创文章,版权所有。转载请注明:转载自Vimer的程序世界 [ http://www.vimer.cn ]

本文链接地址: http://www.vimer.cn/?p=2008

10 个评论 在 “STL可能的误用-find_first_of和erase”

  1. 依云 说:

    那个 erase 的没看懂,我的 C++ 是不行了。。。

    [回复]

    egmkang 回复:

    Effective STL看一遍你就懂了

    PS:
    it = myMap.erase(it);
    也是可以的,erase都有返回迭代器的

    [回复]

    Dante 回复:

    it = myMap.erase(it);
    这种只有微软实现的stl才能用,g++的版本是不会返回的。

    [回复]

    egmkang 回复:

    恩,看了一下map的文档,是不返回的,vector返回的

    [回复]

  2. egmkang 说:

    我貌似还没有在map中连续的删除,所以也没碰到过那个问题….汗

    [回复]

  3. 亚弥 说:

    尽量用++it(在“简洁型”的第二个分支),通常它会比it++有更高的效率。

    [回复]

    Dante 回复:

    编译器会自动替你做这件事情的。

    [回复]

    亚弥 回复:

    不,如果是iterator就不会,因为你可能会为前自加和后自加重载完全不同的函数,虽然不推荐做这个,但是编译器是不能够妄加猜测的。因此编译器在这个时候是不会优化的。

    另外,对于典型的iterator的实现,后自加通常需要保存一份自己的状态,而这显然是会比前自加慢的。

    所以Effective C++里面建议:对内建类型随便,但是对iterator这种尽量采用前自加。而最好的习惯是尽量对所有的情况采用前自加。

    [回复]

    Dante 回复:

    原来这样啊,看来是我搞错了,多谢提醒~

    [回复]

    梧桐 回复:

    这个我前一段时间也看到了,书上说尽量写++i

    [回复]

我要评论

*

*