Paste from Word - Leading space disappears (in formatted block)

[Frédéric Arène]

When I paste from Word formatted text like this one :

<span>A b<b> c d</b></span>

it outputs this in CKEditor :

<p>
	A b<b>c d</b></p>

...with the formatting rendered, it's like that : A b c d -> A bc d

The space betwwen 'b' and 'c' disappeared, (probably somewhere in the htmlWriter).

[Frédéric Arène]

Very simple MSWord document to reproduce the bug.

[Krzysztof Studnik]

Confirmed for 3.4.3 & 3.5

[Krzysztof Studnik]

Does not occur in CKEditor 3.1.1,

[Tomas]

Yet another example of more disappearing spaces.

[Tomas]

Comment added another example of more spaces that disappears when pasting from word. In the example all spaces after words between styling disappear.

trailing space also disappears if first word in document is styled, but the trailing space is not. i.e. <p><b>A<b> B</p> -> <p><b>A<b>B</p> (that is not in the example document.)

[Wiktor Walc]

Check the attached file for one more example (I have used different fonts to reproduce it).

[Tomas]

The problem here is occuring in the function parser.onText() in core\htmlparser\fragment.js

The problem occurs when a span element has been parsed, the code in on text will then call trimLeft, this will cause the space to disappear. This is the code that trims it away (from fragment.js)

if ( !currentNode._.hasInlineStarted && !inPre )
			{
				text = CKEDITOR.tools.ltrimSpecial( text );
				//text = CKEDITOR.tools.ltrim( text );

				if ( text.length === 0 )
					return;
			}

I tried to understand and fix this, but I really dont understand why it is trimmed, I made a local tweak in my installation for this, but it would be great if one of the core developers could have a look at this.

I ended up trimming, if and only if there are more than one space. Ugly fix, but it solves my problem.

[Tomas]

the commented out code is the original code, ltrimSpecial looks like this, if anyone is intrested.

 ltrimSpecial : (function()
		{
			//do not delete a single space, as it probably should be there.
			var trimRegex = /^[ ]{2,}|^[\t\n\r]+/g;
			return function( str )
			{
				return str.replace( trimRegex, '' ) ;
			};
		})(),

[Garry Yao]

@toma25 It's not the parser::onText that is not doing anything wrong as whitespace works well when loading from source.

Proposing context-sensitive parsing as the filter might choose to re-parse the inner html of potentially any element.

[Tomas]

Replying to garry.yao:

@toma25 It's not the parser::onText that is not doing anything wrong as whitespace works well when loading from source.

Proposing context-sensitive parsing as the filter might choose to re-parse the inner html of potentially any element.

Will not argue against you, as I wrote, the code supplied is probably not the right fix, but it did solve my problem. Again, I suggest that one of the developers that understand the code takes a look at this.

[Garry Yao]

@toma25 Thanks for the input, we've taken your patch into consideration, pls trust the team and wait for the ticket to be fixed.

[Garry Yao]

Fixed with [6557].