Adds a release valve for CDATA content checks.

This addresses the issue that some Caja HTML parser clients want to
render content without fixup including content with mismatched
escaping text spans.

The actual implementation of the fixup scheme is in the testcase but
if that looks good overall, it can probably move out of tests into
src.

Submitted @4504

[anupama.dutta, 2011-06-09 10:50:50]

http://codereview.appspot.com/4591043/diff/3001/tests/com/google/caja/parser/...
File tests/com/google/caja/parser/html/NodesTest.java (right):

http://codereview.appspot.com/4591043/diff/3001/tests/com/google/caja/parser/...
tests/com/google/caja/parser/html/NodesTest.java:657: RenderContext
renderContext = new ScriptFixupRenderContext(cat)
If I set withAsciiOnly to false (as we need to do inside
VanillaCajaHtmlSerializer), the renderContext variable no longer remains as an
instance of ScriptFixupRenderContext and the new logic introduced in this patch
in Nodes.java won't get used and an IllegalArgumentException will be thrown as
before.

Please see 
http://codereview.appspot.com/4575056/diff/3/java/gadgets/src/main/java/org/a...
for the shindig change which finally worked for me (but is not acceptable since
we do need to set withAsciiOnly to false) and lines 598 - 604 of
http://codereview.appspot.com/4535111/diff/18001/tests/com/google/caja/parser...
to see how your change can be modified to raise an exception.

[MikeSamuel, 2011-06-09 16:26:00]

http://codereview.appspot.com/4591043/diff/3001/tests/com/google/caja/parser/...
File tests/com/google/caja/parser/html/NodesTest.java (right):

http://codereview.appspot.com/4591043/diff/3001/tests/com/google/caja/parser/...
tests/com/google/caja/parser/html/NodesTest.java:657: RenderContext
renderContext = new ScriptFixupRenderContext(cat)
On 2011/06/09 10:50:50, anupama.dutta wrote:
> If I set withAsciiOnly to false (as we need to do inside
> VanillaCajaHtmlSerializer), the renderContext variable no longer remains as an
> instance of ScriptFixupRenderContext and the new logic introduced in this
patch
> in Nodes.java won't get used and an IllegalArgumentException will be thrown as
> before.
> 
> Please see 
>
http://codereview.appspot.com/4575056/diff/3/java/gadgets/src/main/java/org/a...
> for the shindig change which finally worked for me (but is not acceptable
since
> we do need to set withAsciiOnly to false) and lines 598 - 604 of
>
http://codereview.appspot.com/4535111/diff/18001/tests/com/google/caja/parser...
> to see how your change can be modified to raise an exception.

Quite right.  That was dumb of me.  Will fix.

[anupama.dutta, 2011-06-13 01:17:03]

LGTM.

[Jasvir, 2011-06-13 06:47:04]

LGTM

[Jasvir, 2011-06-14 03:43:23]

While this patch is fine as it stands, as discussed with Mike offline, this
change is in the TCB and would be safer if RenderContext controlled whether
fixup will be applied or not (as in the first patch set) rather than applying it
unconditionally.

[anupama.dutta, 2011-06-14 04:01:31]

On Tue, Jun 14, 2011 at 9:13 AM,  <jasvir@gmail.com> wrote:
> While this patch is fine as it stands, as discussed with Mike offline,
> this change is in the TCB and would be safer if RenderContext controlled
> whether fixup will be applied or not (as in the first patch set) rather
> than applying it unconditionally.
>
> http://codereview.appspot.com/4591043/

Applying this conditionally sounds fine. The first patch set had a
different problem that the way the ScriptFixupRenderContext was
defined and used was not going to go well with the
VanillaCajaHtmlSerializer code that I had pointed out to you - if you
could fix that (basically modify the first patch set so that the
comment/test mentioned in
http://codereview.appspot.com/4591043/diff/3001/tests/com/google/caja/parser/...
is addressed), that should be an acceptable solution?

Thanks,
Anupama.

[Jasvir, 2011-06-15 00:04:37]

LGTM

http://codereview.appspot.com/4591043/diff/17001/src/com/google/caja/parser/h...
File src/com/google/caja/parser/html/Nodes.java (right):

http://codereview.appspot.com/4591043/diff/17001/src/com/google/caja/parser/h...
src/com/google/caja/parser/html/Nodes.java:646: } while (problemIndex != -1);
With the null break above, this appears to terminate however we may want to
consider adding a max counter on this loop.