// C++ search engine

#include <iostream>
#include <fstream>
#include <string>
#include <set>
#include <atlhttp.h>  // only works in Visual C++ 2005 and earlier
					// see http://www.hackthissite.org/articles/read/1078 for ideas on using LibCurl instead
#include <cstdlib>
using namespace std;

set<string> parseWord(string input);

int main()
{
	char web[] = "http://minich.com/education/psu/cplusplus/index.php";
	cout << "input a web site to search: ";
	cin >> web; 
	cin.ignore(100, '\n');

	string input = "";
	cout << "input a term to be searched for: ";
	getline(cin, input);


	CAtlHttpClient client;
    CAtlNavigateData navData;
	client.Navigate(web, &navData );
	const byte *text = client.GetBody();



	bool found = false;
	string word = "";
	set<string> hashTable;


	while (*text != NULL && !found)
	{
		word = "";

		if (*text == '<')
		{
			while (*text != '>')
			{
				text++;
			}
		}
		else
		{
			while (*text != '<')
			{
				word += *text;
				text++;
			}

			cout << word << endl;
			text--;
		}

		hashTable = parseWord(word);

		if (hashTable.count(input) > 0)
		{
			cout << input << " was found on the web page" << endl;
			found = true;
		}

		hashTable.clear();	

		text++;
	}

	if (!found)
	{
		cout << input << " was not found on the web page" << endl;
	}
	
	system("pause");
	return 0;
}// end of main

set<string> parseWord(string input)
{
	set<string> hashTable;

	for (int i = 0; i < input.size(); i++)
	{
		for (int k = input.size() - 1; k >= 0; k--)
		{
			hashTable.insert(input.substr(i, k - i));
		}
	}
	return hashTable;
}

/*
Minich's cleaner original version

#include <iostream>
#include <string>
#include <atlhttp.h>   
using namespace std;

int main()
{
	CAtlHttpClient client;
    CAtlNavigateData navData;
	client.Navigate("http://www.minich.com", &navData );
	const byte *text = client.GetBody();
	string word = "";
	string input = "Chess";
	bool found = false;

	cout << "Enter a word that you'd like to find: ";
	cin >> input;

	while (*text != NULL && !found)
	{
		word = "";

		while (*text != ' ')
		{
			word += *text;
			text++;
		}

		text++;
		
		//cout << word << endl;

		if (word == input)
		{
			cout << word << " was found on the web page" << endl;
			found = true;
		}
	}

	if (!found)
	{
		cout << word << " was not found on the web page" << endl;
	}
	
	return 0;
}
*/