Page MenuHomePhabricator

string "overflow:auto; height:" is recognized as spam
Closed, InvalidPublic

Description

background:
2007-08-25 there was reported a strange behaviour of the spamfilter.[http://de.wikipedia.org/wiki/MediaWiki_Diskussion:Spam-blacklist#Bug.3F] So this bug is _not_ caused by the latest spam blacklist bugfix.

bug:
if one uses the string "overflow:auto; height:" _anywhere_, e.g. in <pre style="overflow:auto; height:4em;"></pre> or just as plaintext, it will be catched by the spamfilter. i tested that in meta and de.

notes:
i'm not sure, whether this is a sbl-bug or a mediawiki-bug. perhaps the link-preparsing "getExternalLinks" is somehow bugged?


Version: unspecified
Severity: enhancement

Details

Reference
bz14141

Event Timeline

bzimport raised the priority of this task from to Medium.Nov 21 2014, 10:10 PM
bzimport set Reference to bz14141.
bzimport added a subscriber: Unknown Object (MLST).

ayg wrote:

Can you confirm that this string is not in fact blacklisted? This isn't really a bug at all if it matches a blacklist entry. In that case, change the blacklist.

This doesn't really look like a bug, and even if it is, it's not with
SpamBlacklist. Have a look at Wikimedia's InitialiseSettings.php
http://noc.wikimedia.org/conf/InitialiseSettings.php.html. $wgSpamRegex is
set to
'/overflow\s*:\s*auto\s*;\s*height\s*:|<div[^>]*font-size[^>]*font-color:\s*transparent[^>]*>/i'
by default, which would most likely explain why the spam filter catches it.

ayg wrote:

This is a Wikimedia config issue, not an issue with the extension. Probably this is to stop spammers who create giant invisible divs that are links to somewhere. Note that the regex is fragile (height: x; overflow: auto; works). And "font-color" is a little weird, since that's not a valid CSS property.

ok, so it's not a bug in wikimedia, but a problem of of the spam protection text in de.wikipedia.org.
thx for quick answers and sorry for my mistake.

Not a bug, this was added to the blacklist config a couple years ago due to a spambot using that particular formulation over and over and over and over...