Issue 6437 - Apply :-abp-contains() to matching roots only

For a filter like div:-abp-contains(Hello) and a subtree like
<div id="a"><div id="b"><div id="c">Hello</div></div></div>
we're only interested in div#a

[hub, 2018-04-08 11:24:47]

this warrant updating the test suite.

[hub, 2018-04-08 11:26:25]

And actually the behaviour we want is only getting div#c (the strictest match)

[Manish Jethani, 2018-04-08 12:09:48]

On 2018/04/08 11:26:25, hub wrote:
> And actually the behaviour we want is only getting div#c (the strictest match)

That would be a change in behavior, wouldn't it? Right now hiding div#a is the
same as hiding div#a, div#b and div#c. If we change this to hide only div#c,
then div#a and div#b will remain visible. I don't mind doing it, I actually
agree with this, but perhaps this should be filed as a separate issue and
discussed? Here's I'm just trying to optimize for the current behavior without
changing the behavior.

[hub, 2018-04-09 03:32:47]

On 2018/04/08 12:09:48, Manish Jethani wrote:
> On 2018/04/08 11:26:25, hub wrote:
> > And actually the behaviour we want is only getting div#c (the strictest
match)
> 
> That would be a change in behavior, wouldn't it? Right now hiding div#a is the
> same as hiding div#a, div#b and div#c. If we change this to hide only div#c,
> then div#a and div#b will remain visible. I don't mind doing it, I actually
> agree with this, but perhaps this should be filed as a separate issue and
> discussed? Here's I'm just trying to optimize for the current behavior without
> changing the behavior.

I was thinking of this yesterday and you a right. It changes the behaviour
though, so we need tests. If the test currently fail its probably because I
failed to provide coverage for the case.

[Manish Jethani, 2018-04-19 16:29:10]

Patch Set 2: Rebase
Patch Set 3: Update tests

The updated tests cover the change nicely. The inner element is not supposed to
have its display set to none.

[Manish Jethani, 2018-04-19 16:29:41]

On 2018/04/09 03:32:47, hub wrote:
> On 2018/04/08 12:09:48, Manish Jethani wrote:
> > On 2018/04/08 11:26:25, hub wrote:
> > > And actually the behaviour we want is only getting div#c (the strictest
> match)
> > 
> > That would be a change in behavior, wouldn't it? Right now hiding div#a is
the
> > same as hiding div#a, div#b and div#c. If we change this to hide only div#c,
> > then div#a and div#b will remain visible. I don't mind doing it, I actually
> > agree with this, but perhaps this should be filed as a separate issue and
> > discussed? Here's I'm just trying to optimize for the current behavior
without
> > changing the behavior.
> 
> I was thinking of this yesterday and you a right. It changes the behaviour
> though, so we need tests. If the test currently fail its probably because I
> failed to provide coverage for the case.

Done.

[hub, 2018-04-19 18:46:28]

LGTM