使用C++ STL库统计一散文中单词出现次数和行号

1、STL容器中的map和vector容器；
2、ifstream库文件流的操作；
3、string的分割查找find、获取子串substr、去除非法字符等待；
4、c++中的基本知识点：构造函数（包括常量的初始化）、引用、对象的构造和析构等等。

/** 
 * @FileName    stl_apply_readfile_1.cpp
 * @Describe    A simple example for using c++ STL to calculate words and line numbers in an article.
 * @Author      vfhky 2017-04-16 16:44 https://typecodes.com/cseries/stlcalcarticlewordlines1.html
 * @Compile 	g++ stl_apply_readfile_1.cpp -o stl_apply_readfile_1
 */
#include <iostream>
#include <map>
#include <vector>
#include <fstream>
#include <algorithm>
#include <string>
#include <assert.h>
#include <string.h>
using namespace std;


class CFileHandle
{
public:
	CFileHandle( const string &s_file_name ) : s_m_file_name(s_file_name)
	{
		i_m_line_number = 0;
	}
	
	//读取每行的数据，然后进行处理
	void ReadFile( const string &s_file_name )
	{
		ifstream inFile;
		
		//以二进制可读的方式打开文件，也可以使用： inFile.open( s_file_name.data() );
		inFile.open( s_file_name.c_str(), ios::in | ios::binary );
		
		//Checks if the file stream has an associated file.
		assert( inFile.is_open() );
		
		//每一行的数据
		string s_line_buf;
		
		//读取一行内容	getline( inFile, s_line_buf )
		while( getline( inFile, s_line_buf, inFile.widen('\n') ) )
		{
			++i_m_line_number;
			
			//统一转换成大写：也可以使用	transform( s_line_buf.begin(), s_line_buf.end(), s_line_buf.begin(), ::toupper );
			transform( s_line_buf.begin(), s_line_buf.end(), s_line_buf.begin(), (int (*)(int))toupper );
			
			//开始数据处理
			HandleLine( s_line_buf );
		}
		
		//关闭文件流
		inFile.close();
	}
	
	//对每行的数据进行解析
	void HandleLine( const string &s_line_buf )
	{
		string s_word_buf;
		int i_start=0, i_last=s_line_buf.size();
		//std::size_t i_split_pos;
		int i_split_pos = 0;
		
		while( i_start < i_last )
		{
			i_split_pos = s_line_buf.find( ' ', i_start );
			string s_word;
			
			//if( i_split_pos == std::string::npos )
			if( i_split_pos == -1 )
			{
				s_word = s_line_buf.substr( i_start, i_last );
				HandleUnkind( s_word );
				HandleWord( s_word );
				break;
			}
			/**
				避免空格，也可以使用下面这两行来去掉行首和行尾的空格
				s_line_buf.erase( 0, s_line_buf.find_first_not_of(" ") );
				s_line_buf.erase( s_line_buf.find_last_not_of(" ") + 1 );
			 */
			else if( i_start != i_split_pos )
			{
				s_word = s_line_buf.substr( i_start, i_split_pos-i_start );
				HandleUnkind( s_word );
				HandleWord( s_word );
				i_start = i_split_pos + 1;
			}
			//对于行首的空格不进行处理
			else
			{
				i_start = i_split_pos + 1;
			}
		}
	}
	
	//去除每个单词可能包含的非字符（除0~9和A~Z外的数据）
	void HandleUnkind( string &s_word_buf )
	{
		assert( s_word_buf.size() );
		
		//char c_word_buf[s_word_buf.size()+1]; 在vs中会报错：error C2131: 表达式的计算结果不是常数
		char *c_word_buf = new char[s_word_buf.size()+1]();
		
		unsigned int j = 0;
		for( unsigned int i=0; i<s_word_buf.size(); ++i )
		{
			if( ( s_word_buf[i] >= 0x30 && s_word_buf[i] <= 0x39 ) || ( s_word_buf[i] >= 0x41 && s_word_buf[i] <= 0x5A ) )
			{
				c_word_buf[j++] = s_word_buf[i];
			}
		}
		
		s_word_buf = c_word_buf;
		delete c_word_buf; 
	}
	
	//对每个单词的处理
	void HandleWord( const string &s_word_buf )
	{
		map<string, vector<int> >::iterator mapit = mapobj.find( s_word_buf );
		
		//如果该单词不存在
		if( mapit == mapobj.end() )
		{
			vector<int> vect;
			vect.push_back( i_m_line_number );
			mapobj.insert( make_pair( s_word_buf, vect ) );
		}
		else
		{
			mapit->second.push_back( i_m_line_number );
		}
	}
	
	//遍历map对象
	void Traverse()
	{
		map< string, vector<int> >::iterator mapit = mapobj.begin();
		cout << "Words\t\t\t\tCounts\t\t\t\tLines" << endl;
		cout << "---------------------------------------------------------------------" << endl;
		for( ; mapit != mapobj.end(); ++mapit )
		{
			cout << mapit->first << "\t\t\t\t" << mapit->second.size() << "\t\t\t\t";
			for( vector<int>::iterator vectorit = mapit->second.begin(); vectorit != mapit->second.end(); ++vectorit )
			{
				cout << *vectorit << " ";
			}
			cout << endl;
		}
	}
	
	//获取文件总行数
	const unsigned int GetTotalLines() const
	{
		return i_m_line_number;
	}
	
private:
	map<string, vector<int> > mapobj;
	unsigned int i_m_line_number;
	const string s_m_file_name;
};


int main( int argc, char **argv )
{
	//文件所在路径
	const string s_file_name =  "STL_APPLY_READFILE_1";
	
	CFileHandle *pCFileHandle = new CFileHandle( s_file_name );
	
	//开始处理文件
	pCFileHandle->ReadFile( s_file_name );
	//打印总行数
	cout << "[" << __FILE__ << ":" << __LINE__ << "] Total Lines=[" << pCFileHandle->GetTotalLines() << "]." << endl;
	//遍历结果
	pCFileHandle->Traverse();
	
	delete pCFileHandle;
	
	return 0;
}