Added handling of $sitekey exceptions

Removed DNT support for better performance; Implemented more or less proper URI parsing; Implemented exceptions via @@*$sitekey=foobar rules (website has to send the same RSA key and a valid signature in X-Adblock-Key header)

[Felix Dahlke, 2012-09-25 12:40:53]

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js
File lib/basedomain.js (right):

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode168
lib/basedomain.js:168: this.spec = spec;
Shouldn't this be this._spec? Seems to be the convention here.

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode169
lib/basedomain.js:169: this._schemeEnd = spec.indexOf(":");
You could probably get rid of a lot of code here if you parse the spec using a
regex with capture groups.

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode173
lib/basedomain.js:173: if (spec[this._schemeEnd + 1] != "/" ||
spec[this._schemeEnd + 2] != "/")
How about this:
if (spec.substr(this._schemeEnd + 1, 2) !== "//")

[Wladimir Palant, 2012-09-25 14:09:41]

I only addressed one of the comments, see below concerning the other ones.

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js
File lib/basedomain.js (right):

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode168
lib/basedomain.js:168: this.spec = spec;
On 2012/09/25 12:40:53, Felix H. Dahlke wrote:
> Shouldn't this be this._spec? Seems to be the convention here.

It's actually a public member like in nsIURI - that's how you get the string
representation of the URI object.

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode169
lib/basedomain.js:169: this._schemeEnd = spec.indexOf(":");
On 2012/09/25 12:40:53, Felix H. Dahlke wrote:
> You could probably get rid of a lot of code here if you parse the spec using a
> regex with capture groups.

Maybe. Then again, that regexp will probably be absolutely unintelligible -
there are just too many optional parts. And there is a slight hope that this
version is actually faster (it does matter here).

http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode173
lib/basedomain.js:173: if (spec[this._schemeEnd + 1] != "/" ||
spec[this._schemeEnd + 2] != "/")
On 2012/09/25 12:40:53, Felix H. Dahlke wrote:
> How about this:
> if (spec.substr(this._schemeEnd + 1, 2) !== "//")

Given that we are testing for string equality here and not whether we actually
deal with the same string instances I'll rather use != operator :)

[Felix Dahlke, 2012-09-25 14:18:31]

> http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode173
> lib/basedomain.js:173: if (spec[this._schemeEnd + 1] != "/" ||
> spec[this._schemeEnd + 2] != "/")
> On 2012/09/25 12:40:53, Felix H. Dahlke wrote:
> > How about this:
> > if (spec.substr(this._schemeEnd + 1, 2) !== "//")
> 
> Given that we are testing for string equality here and not whether we actually
> deal with the same string instances I'll rather use != operator :)

!== doesn't compare instances, it's just like != but doesn't do any type
conversion. I made a habit of always using === and !==, but this is not the
point of my suggestion. This would be equally nice:

[Felix Dahlke, 2012-09-25 14:19:16]

> http://codereview.adblockplus.org/8401061/diff/1/lib/basedomain.js#newcode173
> lib/basedomain.js:173: if (spec[this._schemeEnd + 1] != "/" ||
> spec[this._schemeEnd + 2] != "/")
> On 2012/09/25 12:40:53, Felix H. Dahlke wrote:
> > How about this:
> > if (spec.substr(this._schemeEnd + 1, 2) !== "//")
> 
> Given that we are testing for string equality here and not whether we actually
> deal with the same string instances I'll rather use != operator :)

!== doesn't compare instances, it's just like != but doesn't do any type
conversion. I made a habit of always using === and !==, but this is not the
point of my suggestion. This would be equally nice:

if (spec.substr(this._schemeEnd + 1, 2) != "//")

[Felix Dahlke, 2012-09-25 14:24:26]

LGTM