MediaWiki talk:Usernameblacklist

Does anyone know jack about regular expressions?--Gage 18:30, 5 April 2007 (BST)

Hello. An asterisk means "zero or more of the previous character" in a regular expression. "3*page" would block any user who wanted the string "page" in their name, which you presumably weren't intending. Wikipedia knows about regular expressions, I don't know of any better beginner guides offhand.

I doubt this is going to be useful for anything other than blocking the numerical-username spambots, though - human vandals can just try other names until they hit one that's valid (or read this page to see what characters they need to insert to skirt the restriction). --Kevan 18:37, 5 April 2007 (BST)

Here and here is some information on regular expression in PHP, which should be what mediawiki uses. Also, check wikipedia's Usernameblacklist for pointers on constructing these. The (?i: ) pattern is very important, as it enables case-insensitive matching (I believe). --Brizth 19:03, 5 April 2007 (BST)

Yay Brizth! I was getting Xoid's help with it, and it is 4am his time. I'm totally clueless to tell you the truth :)--Gage 19:16, 5 April 2007 (BST)

Just tell me what you want to blacklist and I'll see what I can do. The second one that's currently on the page should block all names that have only numbers. The first should block all threepage variants. --Brizth 19:25, 5 April 2007 (BST)

(?i:jimbo wales)? You're kidding right?--Vista 19:49, 5 April 2007 (BST)

Copypasta! :D--Gage 19:49, 5 April 2007 (BST)

Don't forget Gold Blade! --^T 03:01, 6 April 2007 (BST)

Testing

For anyone that wants to test a regular expression to make sure you got it right, head to regextester.com, click the button to selected Preg as the dialect (which is what the wiki uses), enter your expression in the first text field, and click the button for enabling the g flag. Then, in the larger text area, enter the list of all the usernames you want to catch (as well as a few you don't want to catch), with each one on a separate line. Ideally, you should see the Result section filled with all the names you didn't want to catch and none of the ones you did want to catch. For more information about the regular expression syntax, either look up a tutorial for PCRE or PREG syntax, or else consult the link provided above if you want some dry reference material. —Aichon— 16:47, 6 September 2012 (BST)

You

As we're doing this, can we throw in uggs? When they turn up, my god are they annoying. --I'm not the Ross UDWiki needs, I'm the Ross it deserves. 09:16, 8 September 2012 (BST)

Can you point me to some examples of the usernames they use? The more, the better. —Aichon— 07:33, 9 September 2012 (BST)

They normally contain the word uggs. Is that something you can specifically block?--I'm not the Ross UDWiki needs, I'm the Ross it deserves. 13:43, 10 September 2012 (BST)

It is, but I'm trying to make the rules as specific as possible, that way we don't accidentally block people with the screen name of "chugg king" or "ugglly bastard" or something. —Aichon— 16:16, 10 September 2012 (BST)

I hear you, uggs as a part probably isnt that common, will run back and have a look. Generally I believe they all begin uggs then random gobbledygook. --I'm not the Ross UDWiki needs, I'm the Ross it deserves. 16:36, 10 September 2012 (BST)

I doubt it's common either, but it's broad enough that I don't want to risk it if possible. If they're all-caps or something, let me know, since we can use that. Linking me to them would be great if you could find, say, 3-5 examples. —Aichon— 16:47, 10 September 2012 (BST)

Roger. --I'm not the Ross UDWiki needs, I'm the Ross it deserves. 16:50, 10 September 2012 (BST)

Suggested blacklists

In this, # = numeral, $ = letter.

• J#i#a$o$ ([1] [2] [3])
• Hanmei#$ ([4] [5] [6] [7])

(I may add more to this list later.) Bob Moncrief ^EBD•W! 13:33, 10 September 2012 (BST)

Added. —Aichon— 16:35, 10 September 2012 (BST)

Current spambot patterns blocked

Expanding on Bob's example above (#=digit, $=letter, ?=either), here are the current patterns I've blocked, hopefully in a more human-readable format for people that want to follow along. All of them are case-insensitive but MUST start with this pattern unless otherwise noted.

$$$$$$### (the $ can be 6-9 long, and the username is not blocked if the ### is 420 or 911)
$#?????##$ (the ? can be 5-7 long)
$#$#$# (does not need to start with it, the $# pattern must be repeated a minimum of 3 times to get caught)
Hanmei#$
Jiaoii##
J#i#a$o$
Jjoggie#ii##
L$$$####$
Linder$#
Lucyer$#
Qi?g#ho$##
Qiao????## (the ? can be 4-7 long)
Uier ##gfi#
Whiaterqu##$
Z###$$$###

I've also made notes where some of the patterns have a range of possibilities (e.g. rather than exactly 5 ? needing to show up, it will catch a range of 5-7 ?). —Aichon— 17:11, 10 September 2012 (BST)

Replacing the Qing one with the Qi one that is just a small bit more inclusive so that it can catch some new variations showing up. —Aichon— 16:16, 11 September 2012 (BST)

I still think we should (continue to) IP block China. It'd be easier if we get Kevan involved, seeing as our IP rangeblocks are pretty lame. ᚱᛁᚹᛖᚾᚨᚾᛏ 13:42, 11 September 2012 (BST)

I'm no longer disagreeing with you on that front. I'm just using this as an additional tool. —Aichon— 15:48, 11 September 2012 (BST)