Sorting multidim array: prioritize if column contains substring, then order by a second column

Personally, I would use a custom (anonymous) function in conjunction with usort().

EDIT: Re – your comment. Hopefully this will put you on the right track. This function gives equal priority to elements which both have EN or neither have EN, or adjusted priority when just one has EN.

usort($array,function ($a, $b) {
    $ac = strpos($a['countries'],'EN');
    $bc = strpos($b['countries'],'EN');
    if (($ac !== false && $bc !== false) || ($ac == false && $bc == false)) {
        return 0;
    }
    elseif ($ac !== false) {
        return 1;
    }
    else {
        return -1;
    }
});

This function, on the other hand, gives equal priority if both have EN, higher if one has EN, and does a text comparison if neither has EN.

usort($array,function ($a, $b) {
    $ac = strpos($a['countries'],'EN');
    $bc = strpos($b['countries'],'EN');
    if ($ac !== false && $bc !== false)) {
        return 0;
    }
    elseif ($ac !== false) {
        return 1;
    }
    elseif ($bc !== false) {
        return -1;
    }
    else {
        if ($a['countries'] == $b['countries']) {
            return 0;
        }
        elseif($a['countries'] > $b['countries']) {
            return 1;
        }
        else {
            return -1;
        }
    }
});

Again, hopefully this will give you enough direction to move forward on your own. If you are having any problems, feel free to post more comments and I’ll try to help. A note if you’re tying to compare multiple properties with weight: try out a funky switch block, e.g.

$ac = array_flip(explode(',',$a['countries']));
$bc = array_flip(explode(',',$b['countries']));
switch (true) {
    case array_key_exists('EN',$ac) && !array_key_exists('EN',$bc):
        return 1;
    case array_key_exists('DE',$ac) && !array_key_exists('EN',$bc) && !array_key_exists('EN',$bc):
        return 1;
    // and so on
}

More Edits!

Actually, I was thinking more on the problem of complex sorting, and I have come up with the following solution, for your consideration. It will allow you to define numerical rankings based on keywords which would appear in the countries index. Here is the code, including an example:

Example Array

$array = array(
    array(
        'countries' => 'EN,DE,SP',
    ),
    array(
        'countries' => 'EN,CH,SP',
    ),
    array(
        'countries' => 'DE,SP,CH',
    ),
    array(
        'countries' => 'DE,SV,SP',
    ),
    array(
        'countries' => 'EN,SP,FR',
    ),
    array(
        'countries' => 'DE,FR,CH',
    ),
    array(
        'countries' => 'CH,EN,SP',
    ),

);

Sorting Routine

$rankings = array(
    'EN' => 10,
    'SP' => 8,
    'FR' => 7,
    'DE' => 5,
    'CH' => 3,
    'SV' => 1,
);
usort($array, function (&$a, &$b) use ($rankings) {
    if (isset($a['_score'])) {
        $aScore = $a['_score'];
    }
    else {
        $aScore = 0;
        $aCountries = explode(',',$a['countries']);
        foreach ($aCountries as $country) {
            if (isset($rankings[$country])) {
                $aScore += $rankings[$country];
            }
        }
        $a['_score'] = $aScore;
    }

    if (isset($b['_score'])) {
        $bScore = $b['_score'];
    }
    else {
        $bScore = 0;
        $bCountries = explode(',',$b['countries']);
        foreach ($bCountries as $country) {
            if (isset($rankings[$country])) {
                $bScore += $rankings[$country];
            }
        }
        $b['_score'] = $bScore;
    }
    if ($aScore == $bScore) {
        return 0;
    }
    elseif ($aScore > $bScore) {
        return -1;
    }
    else {
        return 1;
    }
});

Note: This code will sort the highest ranking entires to the top of the array. If you want reverse behavior, change this:

    elseif ($aScore > $bScore) {

to

    elseif ($aScore < $bScore) {

Note that the greater-than was changed to a less-than symbol. Making this change will result in the lowest ranking entries being sorted to the top of the array. Hope all this helps!

NOTE ALSO!

This code will make a small change to your array, in that it adds the _score element to each array. Hopefully this is not a problem, as by storing this value I was literally able to increase speed by more than double (.00038-.00041 down to .00016-.00018 in my benchmarks). If not, remove the if blocks that retrieve the cached value and let the contents of the else blocks execute every time, except of course for the part which stores the score value.

By the way, here’s a var_export() dump of the array after it was sorted:

array (
  0 => array (
    'countries' => 'EN,SP,FR',
    '_score' => 25,
  ),
  1 => array (
    'countries' => 'EN,DE,SP',
    '_score' => 23,
  ),
  2 => array (
    'countries' => 'EN,CH,SP',
    '_score' => 21,
  ),
  3 => array (
    'countries' => 'CH,EN,SP',
    '_score' => 21,
  ),
  4 => array (
    'countries' => 'DE,SP,CH',
    '_score' => 16,
  ),
  5 => array (
    'countries' => 'DE,FR,CH',
    '_score' => 15,
  ),
  6 => array (
    'countries' => 'DE,SV,SP',
    '_score' => 14,
  ),
)

Enjoy!

Leave a Comment