This was making it through due to a bug in the spam check code. I've updated it and verified these guys are being blocked once again.

v1.20
- Auto-Whitelist for Self, Friends, Group Members and Guild Members.
- Modified Patterns, removed a pattern that would catch too many non-spammers.
- Modified Scrubbers, added a large utf8 scrub list based on a modification of Kyle Smith's "utf8.lua".

The major change on this release is the introduction of "automatic Whitelisting" which will ensure that you will never block: yourself, your friends, anyone you are 'currently' grouped with and anyone seen in any of your guilds.

Additionally I removed a bad pattern that was catching a lot of non-spammers, and re-introduced those which were removed until the Whitelist support had been added.

As usual, enjoy, and let me know if you have any problems.

v1.18
- New Spam Patterns.
- Sliding Expiry for 'Flooding' is only extended when user continues flooding same text, different text does not extend expiry.
- Simplified Scrubbers and Aggregation.
- Normalization output now shows Player/Character Name of sender.
- Fix a bug with monster names not writing out to chat.

I've spent a few hours testing and made several changes to scrubbers, patterns and related code to nearly eliminate the incidence rate for flagging normal users as spammers (i was able to verify this with a couple of my friends.. basically had them go nuts on the keyboard for an hour until they could break it.)

As it turns out, in an attempt to catch some of the multi-line spammers from a few days ago the text scrubbing and normalization got a bit over-zaelous and turned what would normally have been "alphabet soup" into something that resembled a spammer. Whoops.

If this new version appears stable I'll move on to implementing white-listing for friends/etc.

Enjoy!

I humbly apologize, as you say it is in fact "our bug", with the additional info I was able to compare code between the two add-ons and quickly saw the mistake. It is now fixed (v1.18 - which is still pending mod review as I write this.)

Thank you for taking the time to verify that it wasn't being caused by other add-ons, my pre-emptive fear was that you had a number of Add-Ons loaded and had no idea which was causing it and were looking for someone to figure it out. I may have misread.

Again, thanks for the report.

UPDATE: Was able to dig up your screenshot from my browser cache:



I noticed that the add-on isn't actually blocking this message, I didn't notice this before. There are two outputs shown here. The first is 'LibAntiSpam normalization' output which is used for testing/reporting problems, and when the [DEV] feature is enabled this is is displayed for every message. The second line is the actual chat message (not blocked by the Add-On.)

Normally if the Add-On blocks a message as spam:

1. (Optionally, but enabled by default) LibAntiSpam will display a message stating that Spam was detected for a user (and also prints the username, to facilitate reporting spammers.)
2. Does not actually output the text (blocks it.)

So in the case of a block the 'Say' output would never have appeared (whether due to spam block or due to flood block, this is the case.)

If you find some text that shouldn't be caught, but is, send it to me and I'll get it fixed after work today.

v1.17
- Removed debug lines accidentally left in the code (oops!)
v1.16
- Modified Patterns to be more restrictive.
- Added new spam Patterns.
- Modified Scrubbers to deal with link-based spam.


LibAntiSpam also detects 'Flooding', and performs a temporary block with a sliding expiry. The default sliding expiry is 30 seconds, but can be configured for up to 5 minutes. What this means is that if you submit the same series of text within the expiry period the add-on will mark you as a flooder (not a spammer) and temporarily block your text. After 30 seconds (or up to 5 minutes) of not flooding the same text the Add-On will automatically mark you as NOT a flooder. If you resubmit the same text within the period the block period is extended. This has the effect of you appearing to be permanently blocked, when in fact it's because you're flooding chat with the same text.

My suggestion is to increase the interval for your advertisements to something like 5 minutes.

Alternatively, alternate between two advertisements. Examples:

"Join our guild! We rock!'

"Our guild rocks! You should join us!'

This will prevent you from being blocked as a flooder, but if you're spamming the same thing over and over every minute.. you're technically flooding chat. Again I would urge you to increase your interval to 5 minutes. This was more meant to silence people that spam chat every 30-60s with LFG's, WTS's, etc.

In a future release I'll add an option to disable flooder detection entirely.

The latest update (v1.17) adds support for this. It also adds new definitions to catch a new spammer. Thanks for the report! (NOTE: I've deleted your original post so that this spammer does not receive free advertising here as a result of the screenshot.)

NOTE: v1.17 adds more restrictive pattern definitions that should prevent this from happening for most users. Additionally, some users may confuse a "spam block" (which is permanent) with a "flooder block" which is temporary. If you're flooding chat, the Add-On blocks you temporarily.

If you're "Spam Blocked" the add on will print a notification to chat letting you know (but only for the first time it detects you as a spammer.)

If this happens to you:

1. type /reloadui to fix your state
-optionally, to help fix the root cause-
2. Go to Settings=>X4D AntiSpam and enable the [DEV] options.
3. Close the settings UI, and then repeat in chat whatever it was that you said.

The pattern that blacklisted you should be printed in the chat window, along with a normalized string of what you said. If you can provide both of these to me I will fix the issue (screenshot is fine, if this is a problem then simply provide me with the pattern and the original text you sent and I can do the testing myself.)

This can happen if you're repeating spammer websites or some portion of text (such as making fun of a mispelling of "handwork", which tends to catch non-spammers often.)

I'll make plans to automatically white-list the following players:

1. Self
2. Friends
3. Guild Members
4. Party Members

Obviously it's ideal to me that normal chat does NOT get detected, before implementing 'whitelists' based on the above.

I'll try and have these changes in sometime over the next few days, I work full time and have a wonderful girlfriend both of which take up a lot of my free time my availability M-F winds up being limited. I would still appreciate info about which pattern it was that caught you (assuming you feel it shouldn't have.)

Thanks for the Report!

v1.14
- Modifications to text scrubbers (leaner scans, more efficient)
- Spammer notification now only occurs once per spammer, rather than for every message detected.
- Spammer notifications are again enabled by default to promote reporting by users.
- Fixed a bug with SavedVars being reset for every release, what a horrible API.

My apologies to anyone that had custom patterns, the last 2-3 releases wiped them out. This was due to a misunderstanding of the SavedVars API (it does not do differencing, it simply wipes any existing vars for every new version of an Add-On, which is probably why Add-On authors don't actually use the Version parameter.)

Enjoy.

Well, seems one of the more aggressive spammers finally figured it out. Luckily I've had a few days to think about the problem before they did. Thus, v1.13 has been uploaded. This will put an end to 'most' line-break and multi-line spammers without really requiring us to change definitions.

v1.13
- New Algo for Line-Break Spammers.
- New Algo for Multi-Line Spammers.
- Boot in the Mouth for all Spammers.

Enjoy.

Correct, LibAntiSpam will function without any other Add-On installed and does not actually use any other spam or chat library to do its job. It is also designed to be used as a library by other add-on developers, but to my knowledge nobody else has integrated with it yet. I have a Chat Mod (called X4D Chat) which uses LibAntiSpam as a library, but it's not necessary to use them together for it to work. They are both designed to be used independently, but they will also work together. Hopefully other Chat Mod and Mail Mod developers will adopt LibAntiSpam as a 'de facto' library to help analyze, detect and block spam (without necessarily having to worry about implementing anti-spam algorithms, patterns, etc.)

Thanks for the compliments, my goal is to create a spam add-on that will eventually not require a lot of time to maintain. The definition of new spam patterns is still a little painful, but I do plan on simplifying the process of defining them (so that end-users can enter a domain, like "spammerfoo.com" and have the add-on create an appropriate pattern for them.) Further, I would like to see the add-on sync over trusted channels (such as Guild chat, or perhaps via mail between friend-listed players) to help propagate new pattern definitions. I don't believe Add-Ons have a method of talking to one another, yet, and I really hope ZeniMax changes their stance on this for legitimate/trustworthy add-on developers.