Which is more efficient between str_pos and preg_match?

Question

After this question: Pattern for check single occurrency into preg_match_all

I understand that my pattern must contain only a word per cycle because, in the case reported in that question, I must find "microsoft" and "microsoft exchange" and I can't modify my regexp because these two possibilities are given dinamically from a database!

So my question is: which is the better solution between over 200 preg_match and the same numbers of str_pos to check if a subset of char contains these words?

I'm trying to write the possible code for both solution:

$array= array(200+ values);
foreach ($array as $word)
{
    $pattern='<\b(?:'.$word.')\b>i';
    preg_match_all($pattern, $text, $matches);
    $fields['skill'][] = $matches[0][0];
}

the alternative is:

$array= array(200+ values);
foreach ($array as $word)
{
    if(str_pos($word, $text)>-1)
    {
    fields['skill'][] = $word;
    }
}

Answer 1

strpos is much more fast than preg_match, here is a benchmark:

$array = array();
for($i=0; $i<1000; $i++) $array[] = $i;
$nbloop = 10000;
$text = <<<EOD
I understand that my pattern must contain only a word per cycle because, in the case reported in that question, I must find "microsoft" and "microsoft exchange" and I can't modify my regexp because these two possibilities are given dinamically from a database!

So my question is: which is the better solution between over 200 preg_match and the same numbers of str_pos to check if a subset of char contains these words?
EOD;

$start = microtime(true);
for ($i=0; $i<$nbloop; $i++) {
    foreach ($array as $word) {
        $pattern='<\b(?:'.$word.')\b>i';
        if (preg_match_all($pattern, $text, $matches)) {
            $fields['skill'][] = $matches[0][0];
        }
    }
}
echo "Elapse regex: ", microtime(true)-$start,"\n";


$start = microtime(true);
for ($i=0; $i<$nbloop; $i++) {
    foreach ($array as $word) {
        if(strpos($word, $text)>-1) {
            $fields['skill'][] = $word;
        }
    }
}
echo "Elapse strpos: ", microtime(true)-$start,"\n";

Output:

Elapse regex: 7.9924139976501
Elapse strpos: 0.62015008926392

It's about 13 times faster.