Summing durations associated with each MAC address in a MongoDB result

Question

I have a script which get data from MongoDB in the form of array and show the result. The array has a number of records in the thousands. A sample array is as follows:

array(2) {   ["mac"]=> string(17) "2c:33:7a:10:f8:39" int(1478199995) ["duration"]=> int(5) }

array(2) {   ["mac"]=> string(17) "38:0b:40:ad:03:53" int(1478203338) ["duration"]=> int(3) }

array(2) {   ["mac"]=> string(17) "38:0b:40:ad:03:53" int(1478201111) ["duration"]=> int(7) }

array(2) {   ["mac"]=> string(17) "2c:33:7a:10:f8:39" int(1478206709) ["duration"]=> int(7) }

array(2) {   ["mac"]=> string(17) "38:0b:40:ad:03:53" int(1478202821) ["duration"]=> int(6) }

array(2) {   ["mac"]=> string(17) "2c:33:7a:10:f8:39" int(1478202366) ["duration"]=> int(4) }

array(2) {   ["mac"]=> string(17) "38:0b:40:ad:03:53" int(1478205023) ["duration"]=> int(2) } 

I am showing record with respect mac address like in above array MAC address "2c:33:7a:10:f8:39" has three records and "38:0b:40:ad:03:53" has four, so I will add duration of each record and show on browser like this

2c:33:7a:10:f8:39          16sec
38:0b:40:ad:03:53          18sec

I am doing this by following logic which I think is too slow for 5000 to 6000 records.

I first get the MAC address in an array and remove duplication and then run a for loop on that array and add all duration for specific MAC my code is as below.

<?php
 foreach ($cursor as $document) {
 array_push($arr,$document["mac"]);
 }
 $arr=array_unique($arr);
 $duration=0;

 for($i=0;$i<count($arr);$i++){

foreach ($cursor as $document) {

if($document["mac"]==$arr[$i])
{

$duration+=$document['assoc_time'];

}

echo $arr[$i]."      ".$duration;
}

?>

How I can make this process fast and how to perform this task in only one foreach loop.

Answer 1

You can do this in only one foreach, making it faster and intelligible for future modification or reuse. The approach I propose is as follows:

//(some sort of pythonish pseudocode)
// let cursor be the array with the mac addresses and duration
result = [] // empty array
for each element in cursor as document:
  document['mac'] is not a key of result then:
    result[ document['mac'] ] = document['duration']
  otherwise
    result[ document['mac'] ] += document['duration']

This way you can quickly check your data, it is very handy. The only difference is that you'll get an array like this:

[
   'mac1' => 'total duration in integer type',
   'mac2' => 'total duration in integer type for mac2'
   ... etc ..
]

In case you really need the current result array structure, you can do this after the first loop:

//(some sort of pseudocode)
arr = [] //initialize the array
for each element in result as mac => duration:
  arr[] = [ 'mac' => mac, 'assoc_time' => duration ]

The difference between nesting two (possibly) big loops and just checking if a key exists in an array is significant. Now let's write a few lines of PHP code:

<?php
$result = array(); // maybe just $result = []; in case you have a proper PHP version.
foreach ($cursor as $document):
  if ( isset( $result[ $document['mac'] ] ) ):
    $result[ $document['mac'] ] += $document['duration']
  else:
    $result[ $document['mac'] ] = $document['duration']
  endif;
endforeach;

You can just replace the endforeach, endif with the traditional brackets.

Answer 2

You should not be trying to calculate these sums in PHP. Rather, your MongoDB query should use the $sum aggregator to produce exactly the result you want. The MongoDB code would be faster than PHP, and you would be transferring less irrelevant data as well.

Answer 3

Since this is such a simple aggregation use case, I would consider aggregating this information in MongoDB rather than in PHP.

As far as code review comments go, I would state that for a simple section of code, it is kind of a mess stylistically. You have:

• inconsistent indentation
• poor/inconsistent spacing around operators and flow control characters
• inconsistent use of single vs. double quotes in specifying array keys
• poor/inconsistent use of vertical whitespace