void pageRankHelper(myset<WebPage*>& pages, int steps) {
// for each step
for (int i=0; i<steps; i++) {
cout << "STEP " << i+1 << endl;
// create temp map to store PR for current step
map<string,double> temp;
for (set<WebPage*>::iterator it = pages.begin(); it != pages.end(); ++it) {
double sum = 0;
set<WebPage*> in = (*it)->incoming_links();
for (set<WebPage*>::iterator it2 = in.begin(); it2 != in.end(); ++it2) {
sum += (*it2)->PR()/((*it2)->outgoing_links().size() + 1); // all nodes have a self loop
}
// add itself as an incoming AND outgoing link
sum += (*it)->PR()/((*it)->outgoing_links().size() + 1);
temp.insert(pair<string,double>((*it)->filename(), sum));
//cout << " insert:" << (*it)->filename() << " " << sum << endl;
}
// update PR for each page
double total = 0;
for (set<WebPage*>::iterator it = pages.begin(); it != pages.end(); ++it) {
if (temp.find((*it)->filename()) == temp.end()) cout << "CANNOT FIND" << endl;
// update its PR value ----------------------------------make e a variable in config -------
double e = 0.15;
double pr = (1-e)*temp.find((*it)->filename())->second + e/pages.size();
(*it)->setPR(pr);
cout << " " << (*it)->filename() << " " << pr << endl;
total += pr;
}
cout << " " << total << endl;
}
}
void pageRank(myset<WebPage*>& pages, int steps, char* configFile) {
for (set<WebPage*>::iterator it = pages.begin(); it != pages.end(); ++it) {
(*it)->setPR(1.0/pages.size());
}
//pageRankHelper(pages, steps);
}
/**
* Updates the set containing all unique words in the text
*/
void WebPage::all_words(const myset<std::string> & words){
for(myset<std::string> ::iterator it = words.begin(); it != words.end(); ++it){
allWords.insert(*it);
}
}