Case insensitive tags

[goetas]

This should fix #63, but how to apply it only to HTML5 tags, and not to XML tags?

[coveralls]

Coverage remained the same when pulling ca8ed88 on goetas:i63 into f50e879 on Masterminds:master.

[zhaofengli]

Why shouldn't it be applied to XML tags?

[goetas]

Html is case insensitive, xml not (<a>hello<A> is a syntax error)

[zhaofengli]

Does this work? zhaofengli/html5-php@b726aeb

[goetas]

@technosophos what is your opinion about this?

[technosophos]

I think @zhaofengli is on the right track, and I think $untilTag is the right thing to check.

There may be a case in SVG (@mattfarina would know) where the SVG name matches the HTML name, but I don't think that will make a significant different here.

I made a quick suggestion on the patch itself. And we should definitely write some tests to make sure this is working. But otherwise I think we should keep going with that patch.

[zhaofengli]

@technosophos Okay, I've made zhaofengli/html5-php@d505c3f . It's a bit difficult to test the case-sensitive (non-HTML5) scenario at the moment IMO, as <title> and <textarea> are the only elements which are parsed as RCDATA, and both are HTML5 tags. Anyway, the case-insensitive scenario is already tested in @GOETA 's original patch.

[mattfarina]

For reference, SVG is XML. So, it follows XML rules. HTML 5 is where things get different. There is a nice little reference on case sensitivity at http://www.w3.org/TR/html-markup/documents.html#case-insensitivity.

I remember this exact problem when we were originally writing it. I wanted to come back and fix it but never did.

Would it be worth testing that XML cases fail when the case is off? An SVG and MathML test would be useful.

[coveralls]

Coverage increased (+0.01%) when pulling 5f39310 on goetas:i63 into f50e879 on Masterminds:master.

[goetas]

BTW, looking at DOMTreeBuilder implementation, endTag($tagName) does not check anything about the opening tag... essentialy, the opening tag capitalization always wins.

This PR will affect only rcdata tags as textarea or script tags.

@mattfarina If you are interested to have tests about failing test cases in svg or mathml, this will not affect this PR.

[coveralls]

Coverage increased (+0.01%) when pulling 4df8acd on goetas:i63 into b39dd88 on Masterminds:master.