C++ : 自写两个字符串分割函数(含测试程序)

// 文件名 : SplitStrng.h
// 作用   : 分割字符串
#include <vector>
#include <string>
#include <iostream>
#include <algorithm>
using namespace std;

typedef basic_string<char>::size_type S_T;
static const S_T npos = -1;

// split        : 根据分割符严格按照分割符进行分割
// src          : 原始字符串
// delimit      : 为一个字符，严格分割
// null_subst   : 将空串替换为任意多个字符
// 返回分割后的字符串数组
vector<string> split(const string& src, string delimit, string null_subst="")
{
    if( src.empty() || delimit.empty() )
    {
        throw "split:empty string\0";
    }

    vector<string> v;
    S_T deli_len = delimit.size();
    long index = npos, last_search_position = 0;

    while( (index = src.find(delimit, last_search_position)) != npos )
    {
        if(index == last_search_position)
        {
            v.push_back(null_subst);
        }
        else
        {
            v.push_back( src.substr(last_search_position, 
                                index - last_search_position) );
        }

        last_search_position = index + deli_len;  
    }

    string last_one = src.substr(last_search_position);
    v.push_back( last_one.empty()? null_subst:last_one );
    return v;
}

// tokenize     : 根据要过滤的字符分割字符串，并可以选择是否显示空串或替换空串
// src          : 原始字符串
// tok          : 将要过滤掉的字符（如：";," 表示将要过滤掉所有';'和','字符）
// trim         : 指示是否保留空串，默认为保留。
// null_subst   : 将空串替换为任意多个字符
// 返回分割后的字符串数组
vector<string> tokenize(const string& src, 
                        string tok, 
                        bool trim = false, 
                        string null_subst = "")
{  
    if( src.empty() || tok.empty() )
    {
        throw "tokenize: empty string\0";  
    }

    vector<string> v;  
    S_T pre_index = 0, index = 0, len = 0;
    while( (index = src.find_first_of(tok, pre_index)) !=npos )  
    {  
        if( (len = index-pre_index)!=0 )
        {
            v.push_back(src.substr(pre_index, len));
        }
        else if(trim == false)
        {
            v.push_back(null_subst);
        }

        pre_index = index+1;
    }

    string endstr = src.substr(pre_index);
    if( trim == false )
    {
        v.push_back( endstr.empty()?null_subst:endstr );
    }
    else if( !endstr.empty() )
    {
        v.push_back(endstr);
    }

    return v;  
}

//##############################################################################

//测试程序如下：
void SplitStringTest(void)
{
    //----tokenize----test------------------------------------------------------

    string src = ",ab,cde;,fg,,";
    string tok = ";,";

    vector<string> v1 = tokenize(src, tok , true);
    vector<string> v2 = tokenize(src, tok , false, "<null>");

    for(int i = 0; i < v1.size(); i++)
    {
        cout<< "szTmpV1" <<endl;
        string szTmpV1 = v1[i].c_str();
        cout<< szTmpV1 <<endl;
    }

    for(int j = 0; j < v2.size(); j++)
    {
        cout<< "szTmpV2" <<endl;
        string szTmpV2 = v2[j].c_str();
        cout<< szTmpV2 <<endl;
    }

    //----split----test---------------------------------------------------------

    string s = "1\\2\\3\\4";
    string del = "\\";

    try
    {
        vector<string> v3 = split(s, del, "<null>");

        for(UINT k = 0; k < v3.size(); k++)
        {
            cout<< "szTmp" <<endl;
            string szTmp = v3[k].c_str();
            cout<< szTmp <<endl;
        }
    }
    catch (char *s)
    {
        cout<<s<<endl;
    }

    return;
}