状态机学习（一）统计字符

题目来自<系统程序员成长计划>  

作者:李先静.

逐个字符扫描 根据是否为字母决定当前状态  由单词内状态切换到单词外 计数字符加1

#include <iostream>
#include <string>
#include <vector>

using namespace std;

string testString = " this's a test,'Hello World!!!' Welcome to the real world, it sucks, but you are gona love it!!!";

enum STATE {
	STAT_INIT,
	STAT_IN_WORD,
	STAT_OUT_WORD
};

size_t CountWords(const string& s)
{
	int count = 0;
	STATE state = STAT_INIT;
	for (size_t i = 0; i < s.size(); i++)
	{
		switch (state)
		{
		case STAT_INIT:
			if (isalpha(s[i])) {
				state = STAT_IN_WORD;
			}else
			{
				state = STAT_OUT_WORD;
			}
			break;
		case STAT_IN_WORD:
			if (isalpha(s[i])) {
				state = STAT_IN_WORD;
			}
			else {
				count++;
				state = STAT_OUT_WORD;
			}
			break;
		case STAT_OUT_WORD:
			if (isalpha(s[i])) {
				state = STAT_IN_WORD;
			}
			else {
				state = STAT_OUT_WORD;
			}
			break;
		default:
			break;
		}//switch (state)
	}
	if (STAT_IN_WORD == state) {
		count++;
	}
	
	return count;
}

int main()
{
	cout << CountWords(testString) << endl;


       return 0;
}

　　改进版 计数以外还讲单词放入容器

#include <iostream>
#include <string>
#include <vector>

using namespace std;

string testString = " this's a test,'Hello World!!!' Welcome to the real world, it sucks, but you are gona love it!!!";

enum STATE {
	STAT_INIT,
	STAT_IN_WORD,
	STAT_OUT_WORD
};

size_t CountWords(const string& s)
{
	int count = 0;
	STATE state = STAT_INIT;
	for (size_t i = 0; i < s.size(); i++)
	{
		switch (state)
		{
		case STAT_INIT:
			if (isalpha(s[i])) {
				state = STAT_IN_WORD;
			}else
			{
				state = STAT_OUT_WORD;
			}
			break;
		case STAT_IN_WORD:
			if (isalpha(s[i])) {
				state = STAT_IN_WORD;
			}
			else {
				count++;
				state = STAT_OUT_WORD;
			}
			break;
		case STAT_OUT_WORD:
			if (isalpha(s[i])) {
				state = STAT_IN_WORD;
			}
			else {
				state = STAT_OUT_WORD;
			}
			break;
		default:
			break;
		}//switch (state)
	}
	if (STAT_IN_WORD == state) {
		count++;
	}
	
	return count;
}

size_t word_segmentation(const string& s, vector<string>& strVec)
{
	STATE state = STAT_INIT;
	size_t count = 0;
	size_t subStrStart = string::npos;
	size_t subStrEnd = string::npos;

	for (size_t i = 0; i < s.size(); i++)
	{
		switch (state)
		{
		case STAT_INIT:
			if (isalpha(s[i])) {
				subStrStart = i;
				state = STAT_IN_WORD;
			}
			break;
		case STAT_IN_WORD:
			if (!isalpha(s[i])){
				count++;
				subStrEnd = i;
				strVec.push_back(s.substr(subStrStart, subStrEnd- subStrStart));
				state = STAT_OUT_WORD;
			}
			break;

		case STAT_OUT_WORD:
			if (isalpha(s[i])) {
				subStrStart = i;
				state = STAT_IN_WORD;
			}
			break;
		default:
			break;
		}//switch (state)
	}
	
	if (state == STAT_IN_WORD) {
		count++;
		subStrEnd = s.size();
		strVec.push_back(s.substr(subStrStart, subStrEnd- subStrStart));
	}

	return count;
}


int main()
{
	cout << CountWords(testString) << endl;

	vector<string> strVec;
	cout << word_segmentation(testString, strVec) << endl;

	for (vector<string>::iterator it = strVec.begin();
		it != strVec.end(); it++)
	{
		cout << *it << "  ";
	}
	cout << endl;


    return 0;
}

　　运行结果

19
19
this s a test Hello World Welcome to the real world it sucks but y
ou are gona love it
请按任意键继续. . .