Sensitive word replacement algorithm, 4 times more efficient than str_replace (with 6,000 sensitive words attached)

Efficiency comparison (12688 characters, 1 replacement):

• str_replace: 0.109937906265 seconds
  
• strtr: 0.0306839942932 seconds

Comparison of replacement results

• For example: "Zhang San", "Zhang Sanfeng", "Zhang San Toyota" are all prohibited words (Why is there such a distinction? Please see "法X", "法Xgong")
• Original text: "I drove a Honda Toyota to work today"
• strtr: "I drove **** to work today" (replace all four words with *)
• str_replace: "I drove **Toyota to work today" (only the first match was replaced)

So using str_replace to replace cannot essentially solve the problem.


Time comparison:
Number of keywords: 6712 (no duplication) self init:0.00789093971252 (load xcache) self:0.0354378223419 strtr:0.0311169624329 strtr_array:0.432713985443 str_replace:0.109627008438

require('badword.src.php');
$badword1 = array_combine($badword,array_fill(0,count($badword),'*'));
$bb = 'I am open today San Toyota goes to work';
$str = strtr($bb, $badword1);

//Interested friends can study it
function strtr_array(&$str,&$replace_arr) {
 $maxlen = 0;$minlen = 1024*128;
 if (empty($replace_arr)) return $ str;
 foreach($replace_arr as $k => $v) {
 $len = strlen($k);
 if ($len < 1) continue;
 if ($len > $maxlen) $maxlen = $len;
 if ($len < $minlen) $minlen = $len;
 }
 $len = strlen($str);
 $pos = 0;$result = '';
 while ($pos < ; $len) {
 if ($pos + $maxlen > $len) $maxlen = $len - $pos;
 $found = false;$key = '';
 for($i = 0;$i< $maxlen;++$i) $key .= $str[$i+$pos]; //Original text: memcpy(key,str+$pos,$maxlen)
 for($i = $maxlen;$i >= $minlen;--$i) {
 $key1 = substr($key, 0, $i); //Original text: key[$i] = '