On 28 March 2012 13:20, Bináris <wikiposta(a)gmail.com> wrote:
While this was a theoratical problem, I created a
practical one. There are
characters with a shorter code, such as quotation mark (.22) and parentheses
(.28, .29).
Have a look at this section title:
http://hu.wikipedia.org/wiki/Szerkeszt%C5%91:BinBot/semmi#.22D.C3.A1tum.22:…
You will see that the first two .22's (marked here with red, excuse me if
this causes a problem for someone) are encoded quotation marks, while the
last (blue) one a literal .22 as part of a date (Hungarian date order is
yyyy. mm. dd.). I simply don't see any chance to make the difference by bot
unless searching for all section titles in question (as well as anchor
templates) and try to make a reverse match. So this is something very easy
to spoil and almost hopeless to correct.
Another example is listed in the bug report at:
http://sourceforge.net/tracker/?func=detail&group_id=93107&atid=603…
- #802.11n becomes Page{[[Page#IEEE 802n]]} (because \x11 is a
non-printable character).
however: I think we can might be able to work around these two
problems as only characters outside of ASCII are escaped /and/ it has
to be a correct UTF-8 string. Again: check the mediawiki source.
Best,
Merlijn