lexer_v3.l patch removing slowdown in the rule decoding urlencoded characters

[Krzysztof Foltman]

This patch removes a bug that caused extreme slowdown on particular HTML 
code. An example email exploiting this bug is available on request.

I think it's not the only place with this kind of security holes in the 
scanner code (ie. scanner rules that allow specially prepared emails to 
make lexer slow down several thousand-fold), and think it warrants a bit 
of investigation of the scanner.

Krzysztof Foltman
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: excess-lf.diff
URL: <https://www.bogofilter.org/pipermail/bogofilter-dev/attachments/20041217/4dfccfc6/attachment.ksh>