main.cpp

我们一个小组

#include <hash_map>
#include <fstream>
#include <string>
#include <Windows.h>
#include <iostream>
#include <algorithm>

using namespace std ;
using stdext::hash_map ;

class Item
{
public:
	string py_ ;
	wstring word_ ;
	double frequency_ ;

	Item(string py, wstring word, double frequency):py_(py), word_(word), frequency_(frequency)
	{

	}
	bool operator<(Item b)
	{
		bool result ;
		
		if (strcmp(py_.c_str(), b.py_.c_str()) == 0)
		{
			result = ( frequency_ > b.frequency_ ) ;
		}
		else
		{
			result = strcmp(py_.c_str(), b.py_.c_str())<0 ;
		}

		return result ;
	}

	bool operator == (Item b)
	{
		return word_ == b.word_ ;
	}
};


void main()
{
	typedef vector<string> PYTable ;
	hash_map<wstring, PYTable> word2py ;

	ifstream input1("..\\py.txt") ;

	string py ;
	wstring word ;
	string temp ;

	while(input1 >> py)
	{
		input1 >> temp ;

		DWORD dwNum = MultiByteToWideChar (CP_ACP, 0, temp.c_str(), -1, NULL, 0);

		wchar_t *pwText = new wchar_t[dwNum];
		MultiByteToWideChar (CP_ACP, 0, temp.c_str(), -1, pwText, dwNum);

		for (size_t i = 0; i<dwNum-1; i++)
		{
			word = pwText[i] ;
			word2py[word].push_back(py) ;
		}
		
	}
	input1.close() ;



	ifstream input2("..\\fre.txt") ;
	if( !input2.is_open() )
	{
		cout << "error" << endl;
	}
	double frequency ;

	vector<Item> py_word_frequency_table ;

	getline(input2, temp) ;	//ignore the first line
	while (input2 >> temp)
	{
		DWORD dwNum = MultiByteToWideChar (CP_ACP, 0, temp.c_str(), -1, NULL, 0);

		wchar_t *pwText = new wchar_t[dwNum];
		MultiByteToWideChar (CP_ACP, 0, temp.c_str(), -1, pwText, dwNum);
		word = pwText[0] ;

		input2 >> frequency >> frequency >> frequency ;

		hash_map<wstring, PYTable>::iterator it = word2py.find(word) ;
		if ( it!=word2py.end() )
		{
			PYTable& pyTable = it->second ;

			for (PYTable::iterator itr = pyTable.begin(); itr!=pyTable.end(); itr++)
			{
				Item item(*itr, word, frequency) ;
				py_word_frequency_table.push_back(item) ;
			}
		}

		input2 >> frequency ;
	}
	input2.close() ;

	//add 0 frequency word
	for(hash_map<wstring, PYTable>::iterator hash_it = word2py.begin(); hash_it!=word2py.end(); hash_it++)
	{
		Item tempItem("", hash_it->first, 0) ;
		vector<Item>::iterator it = find(py_word_frequency_table.begin(), py_word_frequency_table.end(), tempItem);
		if (it == py_word_frequency_table.end())
		{
			PYTable& t = hash_it->second ;
			for (size_t i = 0; i!=t.size(); i++)
			{
				py_word_frequency_table.push_back(Item(t[i], hash_it->first, 0.0)) ;
			}
		}
	}

	sort(py_word_frequency_table.begin(), py_word_frequency_table.end()) ;
	

	ofstream output("..\\py_word_fre_table.txt") ;
	for (size_t i = 0; i<py_word_frequency_table.size(); i++)
	{
		Item& item = py_word_frequency_table[i] ;

		DWORD dwNum = WideCharToMultiByte(CP_OEMCP, NULL, item.word_.c_str(), -1, NULL, 0, NULL, FALSE);
		char *psText;
		psText = new char[dwNum];
		WideCharToMultiByte (CP_ACP, NULL, item.word_.c_str(), -1, psText, dwNum, NULL, FALSE);
		string w(psText) ;

		delete []psText;

		output << item.py_ << " " << w<< " " << item.frequency_ << endl;
	}
	output.close() ;
}