Distinguish valueless attribs (checked, selected) in the HTML attribute schemas

HTML allows some valueless attributes.
  <input checked>
is equivalent to
  <input checked="checked">

The HTML schema table distinguishes these by putting the attribute
name as the only allowed value.

This change modifies the perl script that derives the JSON doc, and the
JSON HTML attributes, and it adds programmatic support via the HTML.Attrib
class.

It also changes the value criterion used to check untyped HTML attributes
values to allow the value criteria to be converted to Javascript regular
expressions so that they can be evaluated client side.


Submited @3658

[metaweta, 2009-08-26 21:45:41]

LGTM once these nits are fixed.

http://codereview.appspot.com/111042/diff/1/4
File src/com/google/caja/lang/html/HTML.java (right):

http://codereview.appspot.com/111042/diff/1/4#newcode186
Line 186: /** Is the attribute required on all elements on which it is allowed.
*/
End questions with '?'.

http://codereview.appspot.com/111042/diff/1/2
File src/com/google/caja/lang/html/HtmlDefinitions.java (right):

http://codereview.appspot.com/111042/diff/1/2#newcode173
Line 173: // This mapping to integer mapping must stay the same, or version skew
Did you mean "type-to-integer mapping"?

http://codereview.appspot.com/111042/diff/1/2#newcode183
Line 183: A_TYPE_MAP.put(HTML.Attribute.Type.GLOBAL_NAME, 7);
Duplicate entry

http://codereview.appspot.com/111042/diff/1/8
File src/com/google/caja/lang/html/HtmlSchema.java (right):

http://codereview.appspot.com/111042/diff/1/8#newcode115
Line 115: }
indentation

http://codereview.appspot.com/111042/diff/1/8#newcode116
Line 116: }
indentation

http://codereview.appspot.com/111042/diff/1/6
File src/com/google/caja/lang/html/RegularCriterion.java (right):

http://codereview.appspot.com/111042/diff/1/6#newcode92
Line 92: // Use a forward reference to require both to match.
The pattern '?=' is a lookahead assertion, not a forward reference.  A forward
reference is something like
   /(\2two|(one))+/
matching 'oneonetwo'.

[MikeSamuel, 2009-08-26 22:02:19]

http://codereview.appspot.com/111042/diff/1/4
File src/com/google/caja/lang/html/HTML.java (right):

http://codereview.appspot.com/111042/diff/1/4#newcode186
Line 186: /** Is the attribute required on all elements on which it is allowed.
*/
On 2009/08/26 21:45:41, metaweta wrote:
> End questions with '?'.

Done.

http://codereview.appspot.com/111042/diff/1/2
File src/com/google/caja/lang/html/HtmlDefinitions.java (right):

http://codereview.appspot.com/111042/diff/1/2#newcode173
Line 173: // This mapping to integer mapping must stay the same, or version skew
On 2009/08/26 21:45:41, metaweta wrote:
> Did you mean "type-to-integer mapping"?

Done.

http://codereview.appspot.com/111042/diff/1/2#newcode183
Line 183: A_TYPE_MAP.put(HTML.Attribute.Type.GLOBAL_NAME, 7);
On 2009/08/26 21:45:41, metaweta wrote:
> Duplicate entry

Where?

http://codereview.appspot.com/111042/diff/1/8
File src/com/google/caja/lang/html/HtmlSchema.java (right):

http://codereview.appspot.com/111042/diff/1/8#newcode115
Line 115: }
On 2009/08/26 21:45:41, metaweta wrote:
> indentation

Done.

http://codereview.appspot.com/111042/diff/1/8#newcode116
Line 116: }
On 2009/08/26 21:45:41, metaweta wrote:
> indentation

Done.

http://codereview.appspot.com/111042/diff/1/6
File src/com/google/caja/lang/html/RegularCriterion.java (right):

http://codereview.appspot.com/111042/diff/1/6#newcode92
Line 92: // Use a forward reference to require both to match.
On 2009/08/26 21:45:41, metaweta wrote:
> The pattern '?=' is a lookahead assertion, not a forward reference.  A forward
> reference is something like
>    /(\2two|(one))+/
> matching 'oneonetwo'.

Done.

[metaweta, 2009-08-26 22:09:37]

http://codereview.appspot.com/111042/diff/1/2
File src/com/google/caja/lang/html/HtmlDefinitions.java (right):

http://codereview.appspot.com/111042/diff/1/2#newcode183
Line 183: A_TYPE_MAP.put(HTML.Attribute.Type.GLOBAL_NAME, 7);
On 2009/08/26 22:02:19, MikeSamuel wrote:
> On 2009/08/26 21:45:41, metaweta wrote:
> > Duplicate entry
> 
> Where?
> 

Weird.  When I was reviewing it, this line appeared twice.  Never mind.