mb_eregi_replace() expects parameter 3 to be string, null given

[repat]

Pretty odd, I can't find that in your code, since it's only in Dom.php@ clean() and $str has to be a string 🤔 any ideas?

• Version 2.1
• PHP 7.3.8

[paquettg]

Hey @repat ,

mb_eregi_* methods can return false/null if it encounters an error of some kind. I am not able to reproduce this as of yet.

Can you please provide me with the full stack trace and some example html that is causing this to happen for you?

Thank you.

[luyi61]

Same problem. It happened when I loaded a url and the site meta charset is GBK.

Site Url: "https://zhidao.baidu.com/question/521824455220270725.html"

[maksa988]

Same problem, site URL: https://forinsurer.com/ratings/banks/17/3/14

[emulsion-io]

I have same problem,

[MindFlare56]

It looks like this happen when the HTML is gzipped.
It gave me the same problem with amazon.com and that's what I found out searching for a solution.

Content-Encoding: gzip

you can check if a string is gzipped with the following code as suggested by this post
https://stackoverflow.com/questions/10975775/how-to-determine-if-a-string-was-compressed
with the following code
$is_gzip = 0 === mb_strpos($str, "\x1f" . "\x8b" . "\x08", 0, "US-ASCII");

Solution
Add this to the start of Dom.php clean method:

$is_gzip = 0 === mb_strpos($str, "\x1f" . "\x8b" . "\x08", 0, "US-ASCII");
if ($is_gzip) {
    $str = gzdecode($str);
}

solution source: https://www.php.net/manual/fr/function.gzdecode.php

[emulsion-io]

I have the problem on the same site personally but between different URLs that display the same data structure yet

I fix for my case DOM::Clean
I add utf8_encode($str) without the $str returned is NULL

$str = mb_eregi_replace("'\s+>", "'>", utf8_encode($str));
$str = mb_eregi_replace('"\s+>', '">', $str);

[paquettg]

Added code recommended by @AigleBlanc. Fix will be released as part of 2.1.1

[ozonar]

I see this error (after fix) if opened file in Windows-1251 encoding.