#include<stdio.h>

#include <errno.h>
#include <stdlib.h>
#include <string.h>

struct regression {
  long long SX;
  long long SY;
  long long SXX;
  long long SYY;
  long long SXY;
};

struct point {
  int x;
  int y;
};

#include "linear_regression_sample_datasets.h"

void print_linear_regression_results(struct regression* r, int nb_points) {
  printf("SX=%lld\n", r->SX);
  printf("SXX=%lld\n", r->SXX);
  printf("SY=%lld\n", r->SY);
  printf("SXY=%lld\n", r->SXY);

  double a, b;

  b = ((double)nb_points*r->SXY - r->SX*r->SY)/((double)nb_points*r->SXX - r->SX*r->SX);
  a = ((double)r->SY - b*r->SX)/nb_points;

  printf("\nEquation of best fit is: y = %0.4f + %0.4fx\n",a,b);
}

void linear_regression_update(struct regression* r, struct point* point) {
  r->SX  += (long long) point->x;
  r->SXX += (long long) point->x * point->x;
  r->SY  += (long long) point->y;
  r->SXY += (long long) point->x * point->y;
}

void linear_regression(struct point* points, int n) {
  struct regression r;
  r.SX = 0;
  r.SXX = 0;
  r.SY  = 0;
  r.SYY = 0;
  r.SXY = 0;

  for(int i = 0; i < n; i++) {
    linear_regression_update(&r, &points[i]);
  }

  print_linear_regression_results(&r, n);
}


/* Read the points from f and fill the points array.
 *
 * @param f: the input file
 * @param points: the array to fill
 * @param nb_points_max: the size of the array
 * @return the number of points that were read
 */
int read_points(FILE* f, struct point* points, int nb_points_max) {
  /* TODO */
  int i = 0;
  while(fscanf(f, "%d,%d", &points[i].x, &points[i].y)>0) {
    //    printf("read (%d, %d)\n", points[i].x, points[i].y);
    i++;
  }
  return i;
}


/* Return the size of f */
size_t file_size(FILE* f) {
  fseek(f, 0, SEEK_END);
  size_t ret = ftell(f);
  fseek(f, 0, SEEK_SET);
  printf("file size: %lu\n", ret);
  return ret;
}

/* Read points from a file, and perform a linear regression */
void process_file(char* input_file) {
  /* TODO: open the file */
  FILE* f = fopen(input_file, "r");
  if(!f) {
    fprintf(stderr, "cannot open %s: %s\n", input_file, strerror(errno));
    return;
  }

  /* load the points to the points array */
  size_t size = file_size(f); // size in byte of the file
  /* each line contains at least 4 char. Let's allocation a buffer
     large enough to store all the points*/
  size_t nb_points_max = size / 4;
  struct point * points = malloc(sizeof(struct point) * nb_points_max);
  int nb_points = read_points(f, points, nb_points_max);

  linear_regression(points, nb_points);
  
  fclose(f);
}

void process_sample_dataset() {
  struct point *points = dataset_500;
  int nb_points = 500;
  linear_regression(points, nb_points); 
}


int main(int argc, char**argv) {
  if(argc < 2) {
    printf("usage: %s input_file\n", argv[0]);
    return EXIT_FAILURE;
  }

  /* Uncomment this line to enable loading the dataset from the header
     file instead of reading it from a file */
  //process_sample_dataset();

  process_file(argv[1]);

  return(0);
}