From xavier-list at rhnh.net  Sun Oct 22 01:24:52 2006
From: xavier-list at rhnh.net (Xavier Shay)
Date: Sun, 22 Oct 2006 15:24:52 +1000
Subject: [PATCH] Ugly rewrite and redirects
Message-ID: <453B00A4.6060306@rhnh.net>

Certain pages (GMail, for one), break when passed through Hpricot.
It would be beneficial if one could prevent Hpricot parsing and work off
the raw text.

I call this "ugly_rewrite". Basically, if only ugly_rewrites are
registered for a page, it will never be parsed by hpricot.

Patch against r78:
http://rhnh.svnrepository.com/svn/public/ruby/pgpirate/mousehole_ugly_patch_r78.diff

Usage:
http://rhnh.svnrepository.com/svn/public/ruby/pgpirate/

As soon as I figure out how to run mouseHole's test suite I'll also
upload appropriate test coverage. (Getting following error:

test_set_attr(TestParser):
NoMethodError: undefined method `search' for #<HTree::Doc:0xb7c3f968>
    ./test/test_parser.rb:14:in `test_set_attr'
)

Also in this patch is a small hack that will call the "redirect" method
on a mouseHole script if it exists. This gives the script a chance to
redirect a request to somewhere of its own choosing. Among the myriad of
possibilities this facilitates, I'm using it to do AJAX requests from an
arbitrary site to my camping app running on localhost without hitting
any security restrictions.

(Technically you can do this without redirects, but it requires a bogus
URL to be sent to the remote server, and you can't do it with POSTs).

QUESTION:
How could I incorporate the existing syntax for urls? (+, -, ... * is
pretty ugly). Don't really understand how this bit of meta-goodness works.

Xavier
http://rhnh.net