32
For the www (second) pattern, maybe this:
/(^|[^]_a-z0-9-=\"'\/:\.])www\.((([a-zA-Z0-9\-]*\.){1,}){1}([a-zA-Z]{2,6}){1})((\/([a-zA-Z0-9\-\._\?\,\'\/\\+&%\$#\=~])*)*)/
The replacement must be
"'\\1<a href=\"http://www.\\2\\6\" title=\"www.\\2\\6\" rel=\"external\">'.MyTextSanitizer::truncate( 'www.\\2\\6' ).'</a>'"
For the email (fourth) pattern, maybe this:
"/(^|[^]_a-z0-9-=\"'\/:\.])([-_a-z0-9\'+*$^&%=~!?{}]++(?:\.[-_a-z0-9\'+*$^&%=~!?{}]+)*+)@((?:(?![-.])[-a-z0-9.]+(?<![-.])\.[a-z]{2,6}|\d{1,3}(?:\.\d{1,3}){3})(?::\d++)?)/ei"
The bug were you reference to is at another place (checking for registering). If it uses also a regexp, then this part could be used:
[-_a-z0-9\'+*$^&%=~!?{}]++(?:\.[-_a-z0-9\'+*$^&%=~!?{}]+)*+@(?:(?![-.])[-a-z0-9.]+(?<![-.])\.[a-z]{2,6}|\d{1,3}(?:\.\d{1,3}){3})(?::\d++)?