XF 2.0 Why does this spam get through?

[JamesAus]

I have the following spam phrases setup to try and avoid Chinese spam my forum seem to be getting:



It doesn't seem to be working however with multiple threads like this being created. You can see the yellow highlighted characters match one of the spam phrases.

Does anyone have any ideas? Is it a bug?

 

[ENF]

I ran a few tests and it seems to be working fine with kanji character detection. (Maybe the only difference is I have it sent for manual approval, but that shouldn't matter in this case..)

I just used a snippet from a local news article:

　ロシア国防省が２５日、ウクライナ侵攻作戦の重心を、東部の親露派支配地域の拡大に移す方針を表明したのは、ウクライナ軍の激しい抵抗で自軍の犠牲が予想以上に膨らみ、首都キエフを早々に陥落させる当初のもくろみも外れたためだ。露軍が手詰まり状態を打破するため、生物・化学兵器に手を出す危険性は高まっている。

I pulled a few samples to use as spam phrases and it got flagged as needing review. (Including one test on your sample character, 大.)

 

[JamesAus]

Thanks @ENF - any idea why it wouldn't be working in my case, and what to look for?

 

[ENF]

Thanks @ENF - any idea why it wouldn't be working in my case, and what to look for?

I would step back through the configuration on how it's checking for spam.
Making sure that the user groups are set correctly as to not bypass the spam checks, etc.
I have to get offline now, won't be back for awhile.. hopefully someone else can assist in the meantime.

My guess though, a configuration issue or a conflicting setting, or even and addon.

 

[JamesAus]

I would step back through the configuration on how it's checking for spam.
Making sure that the user groups are set correctly as to not bypass the spam checks, etc.
I have to get offline now, won't be back for awhile.. hopefully someone else can assist in the meantime.

My guess though, a configuration issue or a conflicting setting, or even and addon.

Thanks for replying. The only setting I can think to check is this one that is set corrrectly at No

 

[Xon]

Has the option "Maximum messages to check for spam" been set to zero? the default is 5.

Do you have any 3rd party add-ons which grant access to the site? Ie a custom app or tapatalk?

 

[JamesAus]

Has the option "Maximum messages to check for spam" been set to zero? the default is 5.

Do you have any 3rd party add-ons which grant access to the site? Ie a custom app or tapatalk?

Thanks for replying Xon.

The maximum messages to check for spam is:



The only add-ons I have are these:

 

[Xon]

Can you check that full unicode support enabled (utf8mb4)?
There should be a $config['fullUnicode'] = true; in src/config.php

Thanks @ENF - any idea why it wouldn't be working in my case, and what to look for?

How many posts did they make and have they been edited to include the blocked content? The spam checker only runs 5 good posts, and doesn't prevent a spammer then going back and editing spam content back in.

 

[ENF]

Can you check that full unicode support enabled (utf8mb4)?
There should be a $config['fullUnicode'] = true; in src/config.php

I sort of tested this, but it didn't make any difference.
However, I'm willing to bet that the encoding of the characters in the database is why it's not catching the spam phrases.

But...
I spun up a clean XF2 install without any config except the spam test kanji -- worked as expected.
...so I don't know.

Edit: Forgot to add - I tested the case of posting a harmless post with no banned phrases and then edited the post, adding a banned character, the post was properly held by the spam checker after adding the banned phrase via the edit.

 

[JamesAus]

Can you check that full unicode support enabled (utf8mb4)?
There should be a $config['fullUnicode'] = true; in src/config.php


How many posts did they make and have they been edited to include the blocked content? The spam checker only runs 5 good posts, and doesn't prevent a spammer then going back and editing spam content back in.

Thanks Xon, yes I have:
$config['fullUnicode'] = true;

They made over 50 threads and seem to keep going until we catch them

I sort of tested this, but it didn't make any difference.
However, I'm willing to bet that the encoding of the characters in the database is why it's not catching the spam phrases.

But...
I spun up a clean XF2 install without any config except the spam test kanji -- worked as expected.
...so I don't know.

Edit: Forgot to add - I tested the case of posting a harmless post with no banned phrases and then edited the post, adding a banned character, the post was properly held by the spam checker after adding the banned phrase via the edit.

Damn! It's so strange then given I'm pretty much using an out of the box XenForo

 

[ENF]

Thanks Xon, yes I have:
$config['fullUnicode'] = true;

They made over 50 threads and seem to keep going until we catch them



Damn! It's so strange then given I'm pretty much using an out of the box XenForo

I think I found your site @JamesAus - Immediate recommendation is to put some sort of real Q&A in place. You seem to have at least one question in place related to how the new member found the site. And looking again, I see a 2nd question, which I don't even know the answer to, lol.

Having questions related to the topic of your site will help deter unwanted junk posters. One site I'm involved with has nearly 40 Q&A questions that rotate (just one at a time, different for each registration) and spam is extremely rare. (Using all XF default tools, no additional addons) The humans can still do manual answers and give the answers to the bots, but it at least slows them down. -- The other option is to move to fully manual approval mode for new accounts. Depending on the volume of your new members, that could be cumbersome... but you get granular control over the accounts you permit into the site.

 

[JamesAus]

I think I found your site @JamesAus - Immediate recommendation is to put some sort of real Q&A in place. You seem to have at least one question in place related to how the new member found the site. And looking again, I see a 2nd question, which I don't even know the answer to, lol.

Having questions related to the topic of your site will help deter unwanted junk posters. One site I'm involved with has nearly 40 Q&A questions that rotate (just one at a time, different for each registration) and spam is extremely rare. (Using all XF default tools, no additional addons) The humans can still do manual answers and give the answers to the bots, but it at least slows them down. -- The other option is to move to fully manual approval mode for new accounts. Depending on the volume of your new membwebboukers, that could be cumbersome... but you get granular control over the accounts you permit into the site.

I think you might have the wrong site actually, because I just have one Q&A question - it's difficult coming up with questions that aren't easily googleable but people will know the answer to.

Thanks for those suggestions

 

[ENF]

I think you might have the wrong site actually, because I just have one Q&A question - it's difficult coming up with questions that aren't easily googleable but people will know the answer to.

Thanks for those suggestions

No worries, I only see one real Q&A question. The other is just a fill-in field. (The user title from the screenshot is what I searched on)
But anyway, good luck. I hope you can resolve it. I'm really curious as to why your spam tools aren't catching those characters.

 

[JamesAus]

No worries, I only see one real Q&A question. The other is just a fill-in field. (The user title from the screenshot is what I searched on)
But anyway, good luck. I hope you can resolve it. I'm really curious as to why your spam tools aren't catching those characters.

If I somehow manage to find out why, I'll certainly report back! Thanks again for taking the time to assist.

 

[ENF]

If I somehow manage to find out why, I'll certainly report back! Thanks again for taking the time to assist.

No worries and likewise, If I can reproduce the case example, I'll post back here.