I write something like adblock for ie. It has to parse the page, find unwanted ads and replace them with wanted ads.
The problem i faced is when to parse and what to parse.
My first idea was to do parsing in a temporary mime filter, but in this case there is a lot of "unresolved" code (internal/external scripts that can modify documen and are executed after parsing).
It is also possible to make filter for image/*, load wanted ad myself and return it in filter's Read method but i have to modify ad's "href" so the problem is not solved completely. Also, this approach does not work for activex(flash).
After that i tried to parse DOM from BHO, but in this case unwanted objects are displayed for a while before I can replace them.
So the best soluton whould be to parse DOM while it is being built or before it is rendered. Or may be to receive something likeBEHAVIOREVENT_CONTENTREADY for commom elements.

Any ideas on how to solve described problem?

[1035 byte] By [Cout] at [2008-2-2]