tokenizer

tokenizer.cpp

#include <iostream>
#include <fstream>
#include <string>
#include <cassert>

using namespace std;

class Transformer
{
public:
	string transform(string str)
	{
		if (str.empty())
			return str;
		if (str.at(0) >= 'A' && str.at(0) <= 'Z')
		{
			if (str.at(str.length() - 1) == '.')
			{
				sDot = true;
			}
			return str;
		}
		if (sDot && str.at(0) <= 'z' &&  str.at(0) >= 'a')
		{
			str[0] = str.at(0) - 32;
			sDot = false;
		}
		for (unsigned int i = 1; i < str.length(); i++)
		{
			if (str.at(i) <= 'Z' &&  str.at(i) >= 'A')
			{
				str[i] = str.at(i) + 32;
			}
		}
		if (str.at(str.length() - 1) == '.')
		{
			sDot = true;
		}
		return str;
	}
private:
	static bool sDot;
};

bool Transformer::sDot = false;

class Bufor
{
public:
	Bufor(string str) :
		mData(str)
	{
	}
	string parse ()
	{
		string temp;
		Transformer t;
		string result;
		for (unsigned int i = 0; i < mData.length(); i++)
		{
			switch (mData.at(i)){
			case ' ':
			case '\t':
				result += t.transform(temp) + mData.at(i);
				temp.resize(0);
				break;
			default:
				temp += mData.at(i);
			}
		}
		result += t.transform(temp);
		return result;
	}

private:
	string mData;
};



int main(int argc, char *argv[])
{
	assert(argc >= 2);
	string wypis;
	ifstream plik(argv[1]);
	while (!plik.eof())
	{
		getline(plik, wypis);
		Bufor b(wypis);
		cout << b.parse() << endl;
	}
	plik.close();
	return 0;
}