Laravel group data by odd amount of hours throughout days - php

Question

I'm trying to group some data in my Laravel project by a date format that is a bit different to the norm. I've got a database that and a query that fetches "Uptime Checks" for a user's website based on the period they want to look over, I then need to display this to the user as some kind of timeline.
In order to reduce "noise" in the data (where there may not be enough uptime checks for a given period) I'd like to group all of my results within say a 3 hour period throughout the day, so I'd have all of the data for:
2021-05-02 03:00:00
2021-05-02 06:00:00
2021-05-02 09:00:00
and so on, right now I'm bringing back data by the hour, but not sure how to modify this to achieve the desired outcome
// get the uptime checks for past X hours
$uptimeData = UptimeChecks::where('user_id', 1)
->where('monitor_id', 1)
->where('checked_at', '>=', '2021-05-02 13:00:00')
->where('checked_at', '<=', '2021-05-03 13:00:00')
->orderBy('checked_at', 'asc')
->select('event', 'checked_at')
->get();
$uptimeDataTimeline = $uptimeData->groupBy(function ($item, $key) {
$date = Carbon::parse($item->checked_at);
// group by hour, how can I get say every 3 hours worth of data?
return $date->format('Y-m-d H:00:00');
});
$uptimeDataTimeline = $uptimeDataTimeline->map(function ($checksInPeriod, $key) {
$down = 0;
$up = 0;
$total = 0;
$uptime = 0;
$fill = '#1fc777'; // green
// $checksInPeriod is all of the data for a given hour at the moment
// I need to group by a bigger period, say, every 3 hours
// add our events
foreach ($checksInPeriod as $key => $value) {
$total++;
if (strtolower($value['event']) == 'down') $down++;
if (strtolower($value['event']) == 'up') $up++;
}
// calculate uptime
$uptime = floatval(number_format(round($up / $total, 5) * 100, 2, '.', ','));
// fill colours
if ($uptime < 100) $fill = '#9deab8'; // lighter green
if ($uptime < 99) $fill = '#fbaa49'; // amber
if ($uptime < 98) $fill = '#e0465e'; // red
return [
'total_events' => $total,
'down_events' => $down,
'up_events' => $up,
'uptime' => $uptime,
'fill' => $fill
];
});
Not sure how to modify the groupBy function which returns the format since my understanding is that it's not possible to do that? I'm using Carbon by the way.
Update
I've been digging, and have come across the CarbonInterval feature, which allows me to generate some intervals, and I've tried implementing this, I seem to get an equally spaced time period, but my data is out and doesn't contain all of the data between two intervals (see attached image)
$intervals = CarbonInterval::hours(2)->toPeriod($from, $to);
$uptimeDataTimeline = $uptimeData->groupBy(function ($item, $key) use ($intervals) {
$date = Carbon::parse($item->checked_at);
foreach ($intervals as $key => $interval) {
if ($date->hour == Carbon::parse($interval)->addHours(1)->hour) {
$actualHour1 = Carbon::parse($interval)->hour;
if (strlen($actualHour1) == 1) $actualHour1 = "0$actualHour1";
return $date->format("Y-m-d $actualHour1:00:00");
} else if ($date->hour == Carbon::parse($interval)->addHours(2)->hour) {
$actualHour2 = Carbon::parse($interval)->subHours(2)->hour;
if (strlen($actualHour2) == 1) $actualHour2 = "0$actualHour2";
return $date->format("Y-m-d $actualHour2:00:00");
}
}
return $date->format('Y-m-d H:00:00');
});
For instance, I should be seeing all of the checks for the hours 7 and 8 within the 07 key, but instead I'm seeing data for just one hour (hour 11)?

Answer 1

The best thing to use whenever you need time slice(s) is DateInterval or better CarbonInterval. What they give you is the ability to loop over those slices and do equality/unequlity operation of your sample data this way you can easily organise your data by those time slices to their respective "slots"
Here is an general idea on how to
$intervals = \Carbon\CarbonInterval::hours(3)->toPeriod('2021-05-02 13:00:00', '2021-05-03 13:00:00');
//we get time slots of 3 hours between provided datetimes
foreach ($intervals as $date) {
$dtArr[] = strtotime($date->format('Y-m-d H:i:s')); //we collect those "time markers"
}
$result = [
'first'=> 0,
'second'=>0.
'third'=>0,
'forth'=>0,
'fifth'=>0,
'sixth'=>0,
'seventh'=>0,
'eighth'=>0
]; //array to accumulate your aggregations to correct time slot
foreach ($uptimeData as $sample) {
//loop over sample set
$ordinality = getSlotNo($sample->checked_at); //eg. third
//read the accumulated total in $result and add this too
$result[$ordinality] += 1;
}
function getSlotNo($dt){
$ts = strtotime($dt);
//eg. say greater than or equal to "13:00" but smaller than "16:00" -> You go in first slot
if($ts>=$dtArr[0] && $ts<$dtArr[1]){
//first slot
return 'first';
}
elseif($ts>=$dtArr[1] && $ts<$dtArr[2]){
//eg. say greater than or equal to "16:00" but smaller than "19:00" -> You go in second slot
//second slot
return 'second';
}
elseif($ts>=$dtArr[2] && $ts<$dtArr[3]){
//third slot
return 'third';
}
// and so on
}
UPDATE
Try something like this may be, modify the slot getter to "look ahead" and decide the result
$i=0;
foreach ($intervals as $date) {
$dtArr[] = strtotime($date->format('Y-m-d H:i:s')); //we collect those "time markers"
$result['int_'.$i] = 0;
$i++;
}
//fake data
$uptimeData=collect([
(object)['checked_at'=>'2021-05-03 10:10:00'],
(object)['checked_at'=>'2021-05-03 11:20:00'],
(object)['checked_at'=>'2021-05-03 12:20:00'],
(object)['checked_at'=>'2021-05-03 13:20:00'],
(object)['checked_at'=>'2021-05-03 14:20:00'],
]);
foreach ($uptimeData as $sample) {
//loop over sample set
$ordinalInfo = getSlotNo($sample->checked_at, $dtArr); //eg. third
//read the accumulated total in $result and add this too
if($ordinalInfo['match']){
$result['int_'.$ordinalInfo['index']] += 1;
}
}
/**
* #param $dt
* #return int index in $dtArr this value belongs to
*/
function getSlotNo($dt, $dtArr){
$ts = strtotime($dt);
$info = [];
for($i =0; $i<count($dtArr); $i++){
if(!empty($dtArr[$i+1])){ // if not reached the last item ie. still there's a next
if($ts>=$dtArr[$i] && $ts<$dtArr[$i+1]){
//i'th slot
$info=['match'=>true,'index'=>$i];
break;
}
}else{
// at last item ie. ( $i == count($dtArr)-1 )
if($ts<=$dtArr[$i])
$info=['match'=>true,'index'=>$i];
else
$info=['match'=>false,'index'=>NULL];
}
}
return $info;
}