Get ranges from array of days

Question

I have an array of dates as input in the form "YYYY-mm-dd"=>category_id.0 (category_id counting from 0).

$days =array (
  '2014-10-15' => 49,
  '2014-10-16' => 49,
  '2014-10-17' => 49,
  '2014-10-18' => 0,
  '2014-10-19' => 0,
  '2014-10-20' => 0,
  '2014-10-21' => 0,
  '2014-10-22' => 0,
  '2014-10-23' => 0,
  '2014-10-24' => 0,
  '2014-10-25' => 0,
  '2014-10-26' => 0,
  '2014-10-27' => 0,
  '2014-10-28' => 0,
  '2014-10-29' => 0,
  '2014-10-30' => 0,
  '2014-10-31' => 0,
  '2014-11-01' => 0,
  '2014-11-02' => 0,
  /* MISSING 2014-11-03 */
  '2014-11-04' => 0,
  '2014-11-05' => 0,
  '2014-11-06' => 0,
  '2014-11-07' => 0,
  '2014-11-08' => 0,
  '2014-11-09' => 0,
  '2014-11-10' => 0,
  '2014-11-11' => 0,
  '2014-11-12' => 0,
  '2014-11-13' => 0,
  '2014-11-14' => 0,
  '2014-11-15' => 0,
  '2014-11-16' => 0,
  '2014-11-17' => 0,
  '2014-11-18' => 0,
  '2014-11-19' => 0,
  '2014-11-20' => 0,
  '2014-11-21' => 0,
  '2014-11-22' => 0,
  '2014-11-23' => 0,
  '2014-11-24' => 0,
  '2014-11-25' => 0,
  '2014-11-26' => 0,
  '2014-11-27' => 0,
  '2014-11-28' => 0,
  '2014-11-29' => 0,
  '2014-11-30' => 0,
  '2014-12-01' => 0,
  '2014-12-02' => 0,
  '2014-12-03' => 0,
  '2014-12-04' => 0,
  '2014-12-05' => 0,
  '2014-12-06' => 0,
  '2014-12-07' => 0,
  '2014-12-08' => 0,
  '2014-12-09' => 0,
  '2014-12-10' => 0,
  '2014-12-11' => 0,
  '2014-12-12' => 0,
  '2014-12-13' => 0,
  '2014-12-14' => 0,
  '2014-12-15' => 0,
  '2014-12-16' => 0,
  '2014-12-17' => 0,
  '2014-12-18' => 0,
  '2014-12-19' => 0,
  '2014-12-20' => 0,
  '2014-12-21' => 0,
  '2014-12-22' => 1,
  '2014-12-23' => 3,
  '2014-12-24' => 3,
  '2014-12-25' => 3,
  '2014-12-26' => 3,
  '2014-12-27' => 3,
  '2014-12-28' => 3,
  '2014-12-29' => 50,
  '2014-12-30' => 50,
  '2014-12-31' => 50,
  '2015-12-24' => 2,
  '2015-12-25' => 2,
);

I want to build an array of range of dates grouped by category_id.

For example, category_id 49 has 3 consecutive days from 15 to 17, so the first range should be: array('2014-10-15',2014-10-17',49).

the category_id 0 start from 2014-10-18 and end to 2014-12-21 HOWEVER due the missing day 2014-11-03 I must split the range in two parts:

array('2014-10-18','2014-11-02',0) and array('2014-11-04','2014-12-21',0)

The final output is an array of ranges like this:

Array
(
    [0] => Array
        (
            [0] => 2014-10-15
            [1] => 2014-10-17
            [2] => 49
        )

    [1] => Array
        (
            [0] => 2014-10-18
            [1] => 2014-11-02
            [2] => 0
        )

    [3] => Array
        (
            [0] => 2014-11-04
            [1] => 2014-12-21
            [2] => 0
        )

    [4] => Array
        (
            [0] => 2014-12-22
            [1] => 2014-12-22
            [2] => 1
        )

    [5] => Array
        (
            [0] => 2014-12-23
            [1] => 2014-12-28
            [2] => 3
        )

    [6] => Array
        (
            [0] => 2014-12-29
            [1] => 2014-12-31
            [2] => 50
        )

    [8] => Array
        (
            [0] => 2015-12-24
            [1] => 2015-12-25
            [2] => 2
        )

)

Due the possibility of to have holes in days array, I tought that the best way is to get the first day (in temporary order), then add a day for each cycle. Checking the values, I build the range array.

function get_range_from_days($days) {

    // check if input is empty
    if(count($days) == 0) return array();

    // output to fill
    $ranges = array();

    // order dates keys in order to have the first date in temporary order
    ksort($days);

    // get first and last day
    $firstday = key($days);
    end($days);
    $lastday = key($days);

    // get the type of first day (actually the current day where we looks)
    $current_type = $days[$firstday];

    // using datetime object for easy step of 1 day
    $datetime = new \DateTime($firstday);
    $datetime->setTime(9,0,0); // avoid time problems at midnight (it's needed?)

    // do the first step outside the while
    $datetime->add(new DateInterval('P1D'));

    // store old value of day
    $oldday = $firstday;

    // build the first range
    $ranges[] = array($firstday,null,$current_type);

    while(($day = $datetime->format('Y-m-d')) <= $lastday) {

        // if there are holes, fill it with null
        if(!isset($days[$day])) {
            $days[$day] = null;
        }

        // check if type has changed (=>need new range)
        if(($days[$day] !== $current_type)) {
            $ranges[count($ranges)-1][1] = $oldday;
            $ranges[] = array($day,null,$days[$day]);
            $current_type = $days[$day];
        }

            // store previous day
        $oldday = $day;
            // next day
        $datetime->add(new DateInterval('P1D'));

    }

    // complete the last range
    $ranges[count($ranges)-1][1] = $lastday;

    // remove range of holes
    foreach($ranges as $k=>$range) {
        if(is_null($range[2])) {
            unset($ranges[$k]);
        }
    }

    return $ranges;
}

Any improvements? This looks not so nice.

Answer 1

This is the way I would do it:

function getRangeFromDays($days) {
  // collect dates into catagories
  foreach ($days as $date => $catagory) {
    $temp[$catagory][] = $date;
  }
  // process all dates in each catagory
  foreach ($temp as $catagory => $dates) {
    sort($dates);
    $startDate  = $dates[0];
    $finishDate = array_pop($dates);
    // walk through the dates, breaking at gaps
    foreach ($dates as $key => $date)
    if (($key > 0) && (strtotime($date)-strtotime($dates[$key-1]) > 99999)) {
      $result[] = array($startDate,$dates[$key-1],$catagory);
      $startDate = $date;
    }
    // force the end
    $result[] = array($startDate,$finishDate,$catagory);
  }
  return $result;
}

echo '<pre>';
print_r(getRangeFromDays($days));
echo '</pre>';

I use an intermediate array which makes it easier to get the results because it's already segmented by catagory. All that's then left to do is the date ranges. I turn dates into seconds to see if the gap is bigger than one day (> 99999 seconds).