/////////////////////
//
// This program analyses the frequency at which a particular
// website was visited given the url and the name of an index.dat
// file.
#include<iomanip>
#include<fstream>
#include<iostream>
#include<string>
using namespace std;

int main(){

  ////////////////////
  // Declare variables

  // Stuff for input
  string index_file_name; // file name to be opened
  ifstream index_file; // the stream for the index file
  string target_url; // URL we are interested in

  // Stuff for the computing
  int total_number_of_days = 0;
  int total_number_of_hits = 0;
  int days_with_hits = 0;
  bool hit_today = 0;    
  string current_day = "";
  string type, time, date, domain, file, cache; 
  float percentage;

  // Say what the program does
  cout << "This program computes how often a particular URL is visited" << endl;
  cout << "on a daily basis. The inputs are the index.dat file to be" << endl;
  cout << "tested and the URL of interest. The output is the total" << endl;
  cout << "number of days covered in the file, the number of visits " << endl;
  cout << "to the specified URL, and the percentage of days the" << endl;
  cout << "specified URL was visited." << endl << endl;

  // Input the file name to open
  cout << "Please enter the filename to be investigated: ";
  cin >> index_file_name;
  cout << endl;

  // Open the file or die if it fails
  index_file.open (index_file_name.c_str());
  if (!index_file) {
    cerr << "The file named: " << index_file_name << " was not found." << endl;
    exit(1);
  }

  // Input the url to be found
  cout << "Please enter the URL to be searched for: ";
  cin >> target_url;
  cout << endl;

  // start reading through the index file
  while (index_file >> type >> time >> date >> domain >> file >> cache){
    
    // This was the line I put in to make sure that it was reading correctly
    //  cerr << type << " " <<time << " " <<date << " " <<domain << " " <<file << " " << cache << endl;

    // if we have a new day, take note
    if (current_day != date){
      current_day = date;
      total_number_of_days++;
      hit_today = 0; // make sure we know we haven't seen the target URL today
    }
    
    if (domain == target_url){ // FIX, make this a substring search
      // This was a debugging check
      //cerr << "Hit: " << total_number_of_hits << endl;
      total_number_of_hits++;
      if (!hit_today){
	hit_today = 1;
	days_with_hits++;
      }
    }

  } // end while 


  // Compute the percentage
  cerr << "Visits: " << days_with_hits << "      days: " << total_number_of_days << endl;
  percentage = static_cast<float>(days_with_hits) / total_number_of_days;
  percentage *= 100;

  // Now that we are done, compute the stats
  
  cout << "Total days covered: " << total_number_of_days << endl;
  cout << "Total visits to " << target_url << ": " 
       << total_number_of_hits << endl;
  cout << "Percent of days that " << target_url << " was visited: "
       << fixed << setprecision(2) << percentage << "%" << endl; 

}