C++でsplit

間違って消したので復旧
競技プログラミングでは基本的にboostは使えないのでsplitを作った

〜mysplit〜 istringstreamとback_inserter利用

ホワイトスペースでsplit

std::vector< std::string > mysplit(const std::string &str)
{
	std::istringstream iss(str);
	std::vector< std::string > res;
	//                                                     end-of-streamイテレータ↓
	std::copy( std::istream_iterator< std::string >(iss), std::istream_iterator< std::string >(), std::back_inserter(res) );
  return res;
}

〜mysplit1〜 find利用

文字列デミリタ使用可

std::vector< std::string > mysplit1(const std::string &str, const std::string &delim)
{
	std::vector< std::string > res;	
	size_t current, found, delimlen = delim.size(); //std::string::size_type
	
	for( current = 0; ( found = str.find(delim, current) ) != std::string::npos; current = found + delimlen )
		res.push_back( str.substr( current, found-current ) );
	res.push_back( str.substr( current ) );
	return res;
}

〜mysplit2〜 find_first_of利用

複数文字のデミリタ使用可
mysplit1とは違う

std::vector< std::string > mysplit2(const std::string &str, const std::string &delim)
{
	std::vector< std::string > res;
	size_t current = 0, found;

	for( current = 0; ( found = str.find_first_of(delim, current) ) != std::string::npos; current = found + 1 )
		res.push_back( str.substr( current, found - current ) );
	res.push_back( str.substr( current ) );
	return res;
}

〜mysplit3〜 istringstream利用

デミリタは1文字

std::vector< std::string > mysplit3(const std::string &str, const char delim)
{
	std::vector< std::string > res;
	std::istringstream iss( str );
	std::string segment;
	
	while( std::getline(iss, segment, delim) ) res.push_back( segment );
	return res;
}

〜c_style_split〜 strtok(string tokenize)を利用

破壊的、
文字列デミリタ使用可
static変数を持つ->マルチスレッドに非対応->リエントラント版のstrtok_rがある(非標準ライブラリ)
null要素は読み飛ばされる?

int c_style_split(char *str, const char delim[], char *res[])
{
	//破壊的なのでconst char* strにしてコピーしたものをstrtokしたほうがいいかもしれない
	int len = 0;
	char *segment = strtok( str, delim );
	
	while( segment != NULL ){
		res[len++] = segment;
		segment = strtok(NULL, delim);
	}
	return len;
}

string_split.cpp

int main(){
	
	string original = " \tABC DEFG HIJKL MNOPQ RSTU \t VWXYZ \t";
	char original_c_style[] = " \tABC DEFG HIJKL MNOPQ RSTU \t VWXYZ \t";
	
	vector<string> words = mysplit(original);
	cout << "mysplit \t→";
	for(size_t i = 0; i < words.size(); i++) cout << "[" << words[i] << "]";
	cout<<endl;

	vector<string> words1 = mysplit1(original, " \t");
	cout << "mysplit1\t→";
	for(size_t i = 0; i < words1.size(); i++) cout << "[" << words1[i] << "]";
	cout<<endl;

	vector<string> words2 = mysplit2(original, " \t");
	cout << "mysplit2\t→";
	for(size_t i = 0; i < words2.size(); i++) cout << "[" << words2[i] << "]";
	cout<<endl;

        vector<string> words3 = mysplit3(original, ' ');
	cout << "mysplit3\t→";
	for(size_t i = 0; i < words3.size(); i++) cout << "[" << words3[i] << "]";
	cout<<endl;

	char* words_c_style[100];
	int len = c_style_split(original_c_style, " \t", words_c_style);
	cout << "c_style_split\t→";
	for(int i=0; i < len; i++) cout << "[" << words_c_style[i] << "]";
	cout<<endl<<"broken string = [" << original_c_style << "]" << endl;

        return 0;
}

結果

mysplit         →[ABC][DEFG][HIJKL][MNOPQ][RSTU][VWXYZ]
mysplit1        →[][ABC DEFG HIJKL MNOPQ RSTU][ VWXYZ][]
mysplit2        →[][][ABC][DEFG][HIJKL][MNOPQ][RSTU][][][VWXYZ][][]
mysplit3        →[][   ABC][DEFG][HIJKL][MNOPQ][RSTU][ ][VWXYZ][       ]
c_style_split   →[ABC][DEFG][HIJKL][MNOPQ][RSTU][VWXYZ]
broken string = [       ABC]