#!/usr/bin/env php
<?php
// $Id$

/**
 * @file
 * Runs a simulation to predict the release date of Drupal 7.
 *
 * Usage:
 *   ./release-date.php [mean daily change] [std dev daily change] [todays date] [current critical count] [number of simulations] [percentiles to show (optional)]
 */

// Script name.
array_shift($argv);

if (empty($argv)) {
  print "Usage:\n";
  print "  ./release-date.php [mean daily change] [std dev daily change] [todays date] [current critical count] [number of simulations] [percentiles to show (optional)]\n";
}

$mean = array_shift($argv);
$s = array_shift($argv);
$date = strtotime(array_shift($argv));
$start = array_shift($argv);
$times = array_shift($argv);
$percentiles = empty($argv) ? array('5', '25', '50', '75', '95') : explode(',', array_shift($argv));

$results = array();
for ($i = 0; $i < $times; $i++) {
  $results[] = simulate($mean, $s, $start);
}

$days = array();
$result_counts = array();
foreach ($results as $index => $result_set) {
  $result_counts[$index] = count($result_set);
  foreach ($result_set as $day => $count) {
    $days[$day][] = $count;
  }
}

foreach ($days as $day => $counts) {
  sort($days[$day]);
}

$pdata = array();
foreach ($days as $day => $counts) {
  foreach ($percentiles as $key => $percentile) {
    if (isset($counts[count($counts) - round($times * (100 - $percentile) / 100)])) {
      $pdata[$percentile][] = $counts[count($counts) - round($times * (100 - $percentile) / 100)];
    }
  }
}

for ($row = round(max($pdata[max($percentiles)])) + 3; $row >= 0; $row--) {
  print str_pad($row, 4);
  for ($col = 0; $col <= count($pdata[max($percentiles)]); $col++) {
    foreach ($percentiles as $percentile) {
      if (isset($pdata[$percentile][$col]) && round($pdata[$percentile][$col]) == $row) {
        if ((int)$percentile <= 5 || $percentile >= 95) {
          print '.';
        }
        elseif ($percentile <= 25 || $percentile >= 75) {
          print '+';
        }
        else {
          print '*';
        }
        continue 2;
      }
    }
    print ' ';
  }
  print "\n";
}

print "Day ";
$backspace = 0;
for ($col = 0; $col <= count($pdata[max($percentiles)]); $col++) {
  if ($backspace) {
    $backspace--;
    continue;
  }
  $timestamp = $date + 1 + (60 * 60 * 24 * $col);
  if (date('j', $timestamp) == '1') {
    print date('M 1', $timestamp);
    $backspace += 4;
  }
  elseif (date('j', $timestamp) == '15') {
    print '15';
    $backspace += 1;
  }
  else {
    print ' ';
  }
}
print "\n\n";

$print = 0;
$mod = round($times / 1000);
for ($i = min($result_counts) - 3; $i < max($result_counts) + 3; $i++) {
  print str_pad(date('M j Y', $date + 1 + (60 * 60 * 24 * $i)), 12);
  foreach ($result_counts as $result_count) {
    if ($result_count == $i) {
      if ($print % $mod == 0) {
        print '+';
      }
      $print++;
    }
  }
  print "\n";
}
print "\n\n";

sort($result_counts);

$avg = array_sum($result_counts) / count($result_counts);
$std = 0;
foreach ($result_counts as $result_count) {
  $std += ($avg - $result_count) * ($avg - $result_count);
}
$std = sqrt($std / count($result_counts));
$min = reset($result_counts);
$med = $result_counts[round(count($result_counts) / 2) - 1];
$max = end($result_counts);

foreach (array('avg', 'min', 'med', 'max') as $var) {
  $$var = date('M j Y', $date + 1 + (60 * 60 * 24 * $$var));
}
$std = sprintf('%0.02f days', $std);

print "Avg: $avg\n";
print "Std: $std\n\n";
print "Min: $min\n";
print "Med: $med\n";
print "Max: $max\n\n";


function simulate($mean, $s, $start) {
  $days = array();
  $current = $start;
  // Give up on Drupal after 10 years so as to avoid infinite loops.
  for ($i = 0; $i < 3652.5; $i++) {
    $current += $mean;
    $current += vary($s);
    $current = max($current, 0);
    $days[] = $current;
    if ($current == 0) {
      break;
    }
  }
  return $days;
}

function vary($s) {
  // 100 is sufficient for our purposes.
  $sum = 0;
  for ($i = 0; $i < 100; $i++) {
    $sum += mt_rand() / 100;
  }
  return ($sum - (mt_getrandmax() / 2)) / ((mt_getrandmax() / sqrt(12)) / sqrt(100));
}