Normalize the case of boolean attributes values

[francofaa]

epubcheck flags an error for the following attributes on element iframe:

• allowFullScreen="allowFullScreen"
• allowfullscreen=""
• allow="fullscreen"

Error reads as follows:

ERROR(RSC-005): ./Documents/Work/Validator/Testing Blank Files/S_EPUB_AllowFullScreen_ALL.epub/OEBPS/xhtml/jac_9781319048860_back.xhtml(18,106): Error while parsing file: attribute "allowFullScreen" not allowed here; expected attribute...

("..." omits the enumerated allowed attributes on iframe).

Markup from file here for testing:

<iframe src="http://www.youtube.com/embed/xDMP3i36naA" allowFullScreen="allowFullScreen"></iframe>
<iframe src="http://www.youtube.com/embed/xDMP3i36naA" allowfullscreen=""></iframe>
<iframe src="http://www.youtube.com/embed/xDMP3i36naA" allow="fullscreen"></iframe>

HTML validator allows these, so I propose that this may be a bug in epubcheck.

[mattgarrish]

Are you using the alpha 4.2 release or 4.1.0? If 4.1.0, note that the attributes won't be available until EPUB 3.2 support is integrated.

I'm not able to reproduce this in the alpha, though, as both allow and allowfullscreen are defined for iframe in embed.rnc.

[rdeltour]

I could reproduce on the next/4.2.0 branch. The issue only occurs however with the form allowFullScreen="allowFullScreen"; I suppose it will be the case for all the other boolean attributes. I suppose this is something checked at parsing time by the Nu Html Checker, in order to simplify their schemas.

I'll double-check and add either a parse-time check or fix the schemas where needed. Good catch @francofaa!

[mattgarrish]

allowFullScreen="allowFullScreen";

Isn't that because it has to be allowfullscreen="allowfullscreen"? That syntax works fine for me.

[rdeltour]

Oh no wait, I replied too fast. Actually the issue is that the attribute name is lower-case, it should be allowfullscreen and not allowFullScreen…

[rdeltour]

In any case, boolean attributes accept values that are case-insensitive matches for their canonical name, so we should perform some case normalization at parsing time.