[IE] copy and paste lists from own document leads into invalid lists

[Syslogic]

When copying and pasting lists in the CKEditor lists end up malformed in Internet Explorer (7 and 8).

---

Test setup:

1. Add code below

   <p>
   	Paragraph before</p>
   <ul>
   	<li>
   		First item</li>
   	<li>
   		Second item</li>
   	<li>
   		Third item</li>
   	<li>
   		Fourth item</li>
   </ul>
   <p>
   	Paragraph after</p>
   

2. copy the second and the third item and paste it below the last paragraph.

Now the bullets are gone. Visually it looks like one paragraph with between the two items a <br />

3. Press the Code button. The code definitely doesn't look like what was expected: The items are not in an <ul> tag.

4. Press the Code button again. The document is changed compared to the contents which was in the CKEditor before toggling the Code view.

5. Press the Code button again. The code now represents the end result of the paste action.

---

When pasting a subset of a list, the new pasted data should be wrapped in a <ul> tag in order to create valid code.

[Syslogic]

Start situation

[Syslogic]

Copy and paste list items

[Syslogic]

The generated code after pasting

[Syslogic]

The corrected view after toggling the Code button

[Syslogic]

The code representing the corrected view

[Garry Yao]

This's actually an unfixed case of #5626, ticket tc updated with [6299].

[Garry Yao]

Another patch addressing TCs from #6972.

[Alfonso Martínez de Lizarrondo]

We should check if this also fixes correctly #6912

[Sa'ar Zac Elias]

Put in source:

<dl><dd>a</dd><dt>b</dt></dl>

Toggle to wysiwyg and back to source.
Expected:

<dl>
	<dd>
		a</dd>
	<dt>
		b</dt>
</dl>

Result:

<dl>
	<dd>
		a
		<dl>
			<dt>
				b</dt>
		</dl>
	</dd>
</dl>

[Garry Yao]

Beyond addressing this particular issue, the patch is a simplification rewrite of the parse's DOM constructing logic, all current TCs regard parsing have to be verified.

[Sa'ar Zac Elias]

Insert:

<ul><ol></ul>

Expected result:

<ul>
	<li>
		<ol>
		</ol>
	</li>
</ul>

Actual result:

<ul>
	<li>
		<ol>
		</ol>
	</li>
	<li>
		<ol>
		</ol>
	</li>
</ul>

[Garry Yao]

Nice review, not sure why #3828 TC are not enrolled into the parser dt, I'm moving it now with [6546].

[Sa'ar Zac Elias]

Insert:

<li>1</li>
<dt>2</dt>
<dd>3</dd>

Expected:

<ul>
	<li>
		1</li>
</ul>
<dl>
	<dt>
		2</dt>
	<dd>
		3</dd>
</dl>

Actual:

<dl>
	<dt>
		2</dt>
	<dd>
		3</dd>
</dl>

(List item is left out)

[Garry Yao]

It's quite a tricky case, but we should be able to handle it : two times of invalid fixing.

While the correct expected result is a bit counterintuitive perhaps, but it should be instead, as the dl is invalid inside ul thus are lifted above it.

<dl><dt>2</dt></dl><ul><li>1</li></ul>

[Garry Yao]

Fixed with [6561], thanks Saar for hq review.

[Frederico Caldeira Knabben]

I've added a few TCs for the cases found here with [6641].

One of the tests is not passing because the expectation explained on comment:11 is wrong. The one in comment:10 is definitely what we must have.