#include <iostream>
#include <string.h>
#include <string>
#include <ctype.h>
#include <vector>
#include <ext/hash_map>
#include "utils/type.h"
using namespace std;

#define MAX_WORD_LEN 50
#define MIN_WORD_LEN 3

//a valid word cannot:
//1. exceed length MAX_WORD_LEN
//2. short words with length 1 and 2
//3. both start and end with digits (e.g. contain only digits)
void filter(const char *str, vector<string>& result)
{
	char buf[MAX_WORD_LEN + 1];
	int count=0;
	int len=strlen(str);
	bool overflow = false;
	for(int i=0; i<len; i++)
	{
		if(isalnum(str[i]))
		{
			if(count < MAX_WORD_LEN) buf[count++]=str[i];
			else overflow = true;
		}
		else if(count > 0)
		{
			if(overflow)//ignore overflowed word with unreasonable length
			{
				overflow = false;
				count=0;
			}
			else
			{
				buf[count]='\0';
				if(isalpha(buf[0]) || isalpha(buf[count-1]))
					if(count >= MIN_WORD_LEN) result.push_back(buf);
				count=0;
			}
		}
	}
	if(count>0)
	{
		buf[count]='\0';
		if(isalpha(buf[0]) || isalpha(buf[count-1]))
			if(count >= MIN_WORD_LEN) result.push_back(buf);
	}
}

//***************************************
//Define index class. Used when application will use load index function

class InvIdx
{
	public:

		hash_map<string, vector<int> > inv_index;

		//API
		void insert_content(string & content, int vpos)
		{
			vector<string> results;
			filter(content.c_str(), results);
			for (int i = 0;i < results.size();i++) {
				inv_index[results[i]].push_back(vpos);
			}
		}

		void insert(string content, int vpos)
		{
			inv_index[content].push_back(vpos);
		}

		vector<int>& vpos_list(string key)
		{
			return inv_index[key];
		}
};

//***************************************
char all_one_array[9] = {0,1,3,7,15,31,63,127,-1};//0,1,11,111,1111,11111,111111,1111111,11111111

//Helper functions: operation on bitmap
void setbit(char & bitmap, int bit)		//'bit' should belong to [0,7]
{
	bitmap |= (1 << bit);
}

int getBit(char & bitmap, int bit)		//'bit' should belong to [0,7]
{
    return ((bitmap & (1 << bit)) == 0) ? 0 : 1;
}

bool dominates(char &bitmap1, char &bitmap2)		//question:when to use?
{
	if(bitmap1 == bitmap2) return false;
	return ((bitmap1 | bitmap2) == bitmap1);
}

bool isAllOne(char &bitmap, int num_of_keywords)
{
	return (bitmap == all_one_array[num_of_keywords]);
}