Slicing a time range into parts

Question

First question. Be gentle.

I'm working on software that tracks technicians' time spent working on tasks. The software needs to be enhanced to recognize different billable rate multipliers based on the day of the week and the time of day. (For example, "Time and a half after 5 PM on weekdays.")

The tech using the software is only required to log the date, his start time and his stop time (in hours and minutes). The software is expected to break the time entry into parts at the boundaries of when the rate multipliers change. A single time entry is not permitted to span multiple days.

Here is a partial sample of the rate table. The first-level array keys are the days of the week, obviously. The second-level array keys represent the time of the day when the new multiplier kicks in, and runs until the next sequential entry in the array. The array values are the multiplier for that time range.

[rateTable] => Array
    (
        [Monday] => Array
            (
                [00:00:00] => 1.5
                [08:00:00] => 1
                [17:00:00] => 1.5
                [23:59:59] => 1
            )

        [Tuesday] => Array
            (
                [00:00:00] => 1.5
                [08:00:00] => 1
                [17:00:00] => 1.5
                [23:59:59] => 1
            )
        ...
    )

In plain English, this represents a time-and-a-half rate from midnight to 8 am, regular rate from 8 to 5 pm, and time-and-a-half again from 5 till 11:59 pm. The time that these breaks occur may be arbitrary to the second and there can be an arbitrary number of them for each day. (This format is entirely negotiable, but my goal is to make it as easily human-readable as possible.)

As an example: a time entry logged on Monday from 15:00:00 (3 PM) to 21:00:00 (9 PM) would consist of 2 hours billed at 1x and 4 hours billed at 1.5x. It is also possible for a single time entry to span multiple breaks. Using the example rateTable above, a time entry from 6 AM to 9 PM would have 3 sub-ranges from 6-8 AM @ 1.5x, 8AM-5PM @ 1x, and 5-9 PM @ 1.5x. By contrast, it's also possible that a time entry may only be from 08:15:00 to 08:30:00 and be entirely encompassed in the range of a single multiplier.

I could really use some help coding up some PHP (or at least devising an algorithm) that can take a day of the week, a start time and a stop time and parse into into the required subparts. It would be ideal to have the output be an array that consists of multiple entries for a (start,stop,multiplier) triplet. For the above example, the output would be:

[output] => Array
    (
        [0] => Array
            (
                [start] => 15:00:00
                [stop] => 17:00:00
                [multiplier] => 1
            )

        [1] => Array
            (
                [start] => 17:00:00
                [stop] => 21:00:00
                [multiplier] => 1.5
            )
    )

I just plain can't wrap my head around the logic of splitting a single (start,stop) into (potentially) multiple subparts.

Answer 1

I would use a different approach, and I will change the rateTable representation based of a couple of considerations.

• The $rateTable describe intervals, why don't you encode them properly?
• What happens on the frontiers (Tuesday and Monday in my example use the two different approaches to the boundary definition);
• The results you get are of a comparable type but use a different representation.
• 23:59:59=> seems an hack to me. I can't explain right now but I've a bell ringing in the back of the head telling me to watch out for it.

Last but not least, my personal experience let me say that if you can't wrap your head on an algorithm it is likely that your co-workers will have the same difficulties (even if you succeed and resolve the issues) and the code will be a primary source of bug. If you find a simpler and efficient solution it will be a gain of time, money and headaches. Maybe it will be a gain even if the solution is not so efficient.

$rateTable = array(
    'Monday' => array (
        array('start'=>'00:00:00','stop'=>'07:59:59','multiplier'=>1.5),
        array('start'=>'08:00:00','stop'=>'16:59:59','multiplier'=>1),
        array('start'=>'17:00:00','stop'=>'23:59:59','multiplier'=>1.5)
    ),
    'Tuesday'=> array (
        array('start'=>'00:00:00','stop'=>'08:00:00','multiplier'=>1.5),
        array('start'=>'08:00:00','stop'=>'17:00:00','multiplier'=>1),
        array('start'=>'17:00:00','stop'=>'23:59:59','multiplier'=>1.5)
    )
);

function map_shift($shift, $startTime, $stopTime)
{
    if ($startTime >= $shift['stop'] or $stopTime <= $shift['start']) {
        return;
    }
    return array(
        'start'=> max($startTime, $shift['start']),
        'stop' => min($stopTime, $shift['stop']),
        'multiplier' => $shift['multiplier']
    );
}

function bill($day, $start, $stop)
{
    $report = array();
    foreach($day as $slice) {
        $result = map_shift($slice, $start, $stop);
        if ($result) {
           array_push($report,$result);
        }
    }
    return $report;
}



/* examples */
var_dump(bill($rateTable['Monday'],'08:05:00','18:05:00'));
var_dump(bill($rateTable['Monday'],'08:05:00','12:00:00'));
var_dump(bill($rateTable['Tuesday'],'07:15:00','19:30:00'));
var_dump(bill($rateTable['Tuesday'],'07:15:00','17:00:00'));

At the very least you need a function to convert the original format to the new one.

$oldMonday = array (
   '00:00:00'=>1.5,
   '08:00:00'=>1,
   '17:00:00'=>1.5,
   '23:59:59'=>1
);

function convert($array) 
{
    return array_slice(
        array_map(
           function($start,$stop, $multiplier) 
           {
               return compact('start', 'stop','multiplier');
           },
           array_keys($array),
           array_keys(array_slice($array,1)),
           $array),
        0,
        -1);
}

var_dump(convert($oldMonday));

And yes, you could do the conversion on the fly with

bill(convert($oldRateTable['Tuesday']),'07:15:00','17:00:00');

but if you care a bit of performances...

Answer 2

Here's my method

I converted everything to seconds to make it a lot easier.

Here's the rate table indexed by seconds. Theres only 3 time slots for monday

// 0-28800 (12am-8am) = 1.5
// 28800-61200 (8am-5pm) = 1
// 61200-86399 (5pm-11:50pm) = 1.5

$rate_table = array(
    'monday' => array (
        '28800' => 1.5,
        '61200' => 1,
        '86399' => 1.5
    )
);

It uses this function to convert hh:mm:ss to seconds

function time2seconds( $time ){
    list($h,$m,$s) = explode(':', $time);
    return ((int)$h*3600)+((int)$m*60)+(int)$s;
}

This is the function that returns a rate table

function get_rates( $start, $end, $rate_table ) {

    $day = strtolower( date( 'l', strtotime( $start ) ) );

    // these should probably be pulled out and the function
    // should accept integers and not time strings
    $start_time = time2seconds( end( explode( 'T', $start ) ) );
    $end_time = time2seconds( end( explode( 'T', $end ) ) );

    $current_time = $start_time;

    foreach( $rate_table[$day] as $seconds => $multiplier ) {

        // loop until we get to the first slot
        if ( $start_time < $seconds ) {
            //$rate[ $seconds ] = ( $seconds < $end_time ? $seconds : $end_time ) - $current_time;

            $rate[] = array (

                'start' => $current_time,
                'stop' => $seconds < $end_time ? $seconds : $end_time,
                'duration' => ( $seconds < $end_time ? $seconds : $end_time ) - $current_time,
                'multiplier' => $multiplier

            );

            $current_time=$seconds;
            // quit the loop if the next time block is after clock out time
            if ( $current_time > $end_time ) break;
        }

    }

    return $rate;
}

Here's how you use it

$start = '2010-05-03T07:00:00';
$end = '2010-05-03T21:00:00';

print_r( get_rates( $start, $end, $rate_table ) );

returns

Array
(
    [0] => Array
        (
            [start] => 25200
            [stop] => 28800
            [duration] => 3600
            [multiplier] => 1.5
        )

    [1] => Array
        (
            [start] => 28800
            [stop] => 61200
            [duration] => 32400
            [multiplier] => 1
        )

    [2] => Array
        (
            [start] => 61200
            [stop] => 75600
            [duration] => 14400
            [multiplier] => 1.5
        )

)

Basically the code loops over the rate table and finds how many seconds from the given time slot belong to each rate.