@@ -63,13 +63,13 @@ local zip_patterns = { | |||
local txt_trie | |||
local txt_patterns = { | |||
html = { | |||
{[[(?i)<html\b]], 32}, | |||
{[=[(?i)<html[\s>]]=], 32}, | |||
{[[(?i)<script\b]], 20}, -- Commonly used by spammers | |||
{[[<script\s+type="text\/javascript">]], 31}, -- Another spammy pattern | |||
{[[(?i)<\!DOCTYPE HTML\b]], 33}, | |||
{[[(?i)<body\b]], 20}, | |||
{[[(?i)<table\b]], 20}, | |||
{[[(?i)<a\b]], 10}, | |||
{[[(?i)<a\s]], 10}, | |||
{[[(?i)<p\b]], 10}, | |||
{[[(?i)<div\b]], 10}, | |||
{[[(?i)<span\b]], 10}, |