/* CS 377 Home work 1
   Matthew Harrison 20982469
   <mharriso@student.umass.edu>
   
   So, as always, I procrastinated.  And as always, there are bugs.  Many.  
   You'll notice that my sets are horribly broken - it only records one instance
   of a word in a file.
   Also, this is way longer than it needs to be and is rather a mess in places.
   Next time, I'll start earlier.
   
*/

#include <iostream>
#include <fstream>
#include <string>
#include <map>
#include <set>
#include <iterator>
#include <vector>

//set this to true if you want to see my debugging output
#define DEBUG true

using namespace std;

void debug(string d) {
	if(DEBUG) {
		cout << d << endl;
	}
}

/*
================================================================================
Begin class inverter
================================================================================
*/
class inverter {
	private:
		map<string, set<int> > invertedIndex;
		
	public:
		inverter();
		~inverter();
		void insertWord(string, int);
		void printIndex();
		void printSet(set<int>&);
};


inverter::inverter() {
	debug("New inverter created.");
}

inverter::~inverter() {
	
}

/*search invertedIndex for the word.
  If it exists, add the file # it came
  from into the set.  If it doesn't exist,
  add it to the map.
*/
void inverter::insertWord(string word, int fileIndex) {
	set<int> s;
	s.insert(fileIndex);
	this->invertedIndex.insert(make_pair(word, s));
	//here's where I *tried* to fix my sets
	//map<string, set<int> >::iterator mi = this->invertedIndex.find(word);
	//set<int> si = mi->second;

	
}

/* This cycles through my invertedIndex and prints it out as per the
   instructions.  Notice the lack of proper set functionality.
*/
void inverter::printIndex() {
	map<string, set<int> >::iterator it;
	for(it = this->invertedIndex.begin(); it != this->invertedIndex.end(); ++it) {		
		cout << it->first << ":";
		printSet(it->second);
		cout << endl;
	}
	
}

//Print out a given set, called from printIndex()
void inverter::printSet(set<int> &s) {
	set<int>::iterator it;
	for(it = s.begin(); it != s.end(); ++it) {
		cout << *it << " ";
	}
}
/*
================================================================================
End class inverter
================================================================================
*/

/*
================================================================================
Begin class files
================================================================================
*/

class fileParser {
	private:
		map<int, string> fileList;//this turned out to be useless
		int fileCounter;
	public:
	        fileParser();
                ~fileParser();
                void loadFiles(string&);
		void parseFile(string&, int, inverter&);
		void tokenize(string, int, inverter&);
};


fileParser::fileParser() {
	this->fileCounter = 0;
}

fileParser::~fileParser() {
	
}


//Open the file passed to main() and populate fileList with the files therein
void fileParser::loadFiles(string &flist) {
	ifstream f;
	f.open(flist.c_str());
	debug("Reading from file: " + flist);
	char ch;
	string curFileName;
        inverter inv;
	if(!f.good()) {
		cout << "Error: Input file not valid." << endl;
		exit(-1);
	}
	while(!f.eof()) {
		f.get(ch);
		if(ch != '\n') {
			curFileName += ch;
		} else {
			if(curFileName.size()) { //a bit of a kludge to ignore blank lines
				this->fileList[fileCounter] = curFileName;
				debug("Found file: " + curFileName);
				//parse the current file here
                                parseFile(curFileName, this->fileCounter, inv);
				this->fileCounter++;
				curFileName.clear();
			}
		}
        }
	f.close();
	inv.printIndex();//display the output once all files have been parsed

}

/* This reads in fileName, then sends each line to tokenize()
*/
void fileParser::parseFile(string &fileName, int findex, inverter &inv) {
        ifstream f;
	f.open(fileName.c_str());
	char ch;
	string s;
	while(!f.eof()) {
		f.get(ch);
		//strip these because I'm too lame to understand strtok()
		if((ch != '\n') && (ch != '.') && (ch != '?') && (ch != '!') && (ch != ',') && (ch != '#') && (ch != '@')) {
			s += ch;
		} else {
			if(s.size()) {
				debug("Tokenizing: " + s);
				tokenize(s, findex, inv);
				s.clear();
			}
		}
	}
	f.close();
}


/* This just splits up words based on spaces and adds them to the invertedIndex.
*/
void fileParser::tokenize(string s, int findex, inverter &inv) {
	char *tok = const_cast<char*>(s.c_str());//OMG HAX
	char *result = strtok(tok, " ");

	while(result != NULL) {
		inv.insertWord(result, findex);
		result = strtok(NULL, " ");
	}
}

/*
================================================================================
End class files
================================================================================
*/


void test(string fname) { //method for debugging
	fileParser f;
	f.loadFiles(fname);

}

int main(int argc, char* argv[]) {
	if (argc < 2) {
		cout << "Error:	 Must provide an input file." << endl;
		exit(-1);
	}
	test(argv[1]);
	return 0;
}