The following may provide some hints w/r/t directly deleting malformed pages
from the database:
http://nielsmayer.com/whatididtopoorsql.txt
There are unresolved bugs related to this class of issues:
http://jira.xwiki.org/jira/browse/XE-376
http://jira.xwiki.org/jira/browse/XE-336
IMHO, I think this class of issue should be given higher priority as it is
easy to generate all kinds of system errors just from normal user-activity
-- such as entering "everyday usage" characters in a document title or
document name. It is somewhat scary opening up a wiki to the "unwashed
masses" knowing they can cause errors that require SQL surgery to the
database.
Also, it seems like there should be an "encoding" fix for this, given that
it's possible to even have "nonwestern" chars and text directions as a URL:
http://ar.wikipedia.org/wiki/%D8%A7%D9%84%D9%85%D9%85%D9%84%D9%83%D8%A9_%D8…
See
http://forums.mozillazine.org/viewtopic.php?f=7&t=557289&start=0&am…
details.
Niels
http://nielsmayer.com