HTML Parser: Out of memory

[Frederico Caldeira Knabben]

During some tests with our so loved and weird IE, I was able to, somehow, end up with an innerHTML that I was able to reduce to the following:

<TABLE>
	<TR>
		<TD></TD>
		<P></P>
		<TD></TD>
	</TR>
</TABLE>

I can't recall how I got that HTML (maybe during some block styling task), but the fact is that a user may have such thing while using the editor in IE. The browser simply accepts it.

The big problem is not the above though. The real issue is the, if you try to retrieve the above HTML from the editor, or even feed it into the source view, it will throw you "Out of memory" into the html parser code.

[Frederico Caldeira Knabben]

While coding a fix for this one, I've started writing some tests, to ensure it was not breaking anything. At the end, I found out that there were several cases not passing. I was not anymore able to understand if that was caused by my changes, so I've started fixing all of them, changing the way the parser works a bit. It should be much better now.

[Garry Yao]

The patch is good and passed almost all the regressions，but it failed on some of my un-commited TC and I've provided an appending patch to fix them with the failed TCs, please review it also.

[Frederico Caldeira Knabben]

garry.yao's proposal fix the added tc, but it somehow duplicates the code we have in the "else if" block in that function. Also, it misses cases like <b><i></b></i>.

Also, the forceSimpleAmpersand is a feature of the htmlwriter plugin, not available in the basicWriter, which is well... basic.

I'm coming with a new ticket now.

[Frederico Caldeira Knabben]

While debugging around #3091, I've found out that this new patch still misses a specific case. I'll come with yet another patch for it.

[Garry Yao]

R+ for 3195_4.patch, and I'm again appending a specific new TC and fixing them with the patch based on the reviewed patch.

[Frederico Caldeira Knabben]

Fixed with [3314].