<source> with only spaces are emptied on deserialization

[ysavourel]

It looks like the content of <source> is removed on reading when it is made of spaces. For example if we have this:

String data = "<xliff srcLang='en' version='2.0' xmlns='urn:oasis:names:tc:xliff:document:2.0'>"
    + "<file id='f1'><unit id='u1'>"
    + "<segment><source>Sentence 1.</source></segment>"
    + "<ignorable><source> </source></ignorable>"
    + "<segment><source>Sentence 2.</source></segment>"
    + "</unit></file></xliff>";
using (IO.MemoryStream ms = new IO.MemoryStream(Encoding.UTF8.GetBytes(data)) )
{
    XliffReader reader = new XliffReader();
    XliffDocument doc = reader.Deserialize(ms);
    foreach (XliffElement e in doc.CollapseChildren<XliffElement>() )
    {
        Console.WriteLine("Type: " + e.GetType().ToString());
        if ( e is PlainText )
        {
            PlainText pt = (PlainText)e;
            Console.WriteLine("Content: '" + pt.Text + "'");
        }
    }
}

We get this output (no content for <ignorable>):

Type: Localization.Xliff.OM.Core.File
Type: Localization.Xliff.OM.Core.Unit
Type: Localization.Xliff.OM.Core.Segment
Type: Localization.Xliff.OM.Core.Source
Type: Localization.Xliff.OM.Core.PlainText
Content: 'Sentence 1.'
Type: Localization.Xliff.OM.Core.Ignorable
Type: Localization.Xliff.OM.Core.Source
Type: Localization.Xliff.OM.Core.Segment
Type: Localization.Xliff.OM.Core.Source
Type: Localization.Xliff.OM.Core.PlainText
Content: 'Sentence 2.'

While I would expect this output:

Type: Localization.Xliff.OM.Core.File
Type: Localization.Xliff.OM.Core.Unit
Type: Localization.Xliff.OM.Core.Segment
Type: Localization.Xliff.OM.Core.Source
Type: Localization.Xliff.OM.Core.PlainText
Content: 'Sentence 1.'
Type: Localization.Xliff.OM.Core.Ignorable
Type: Localization.Xliff.OM.Core.Source
Type: Localization.Xliff.OM.Core.PlainText
Content: ' '
Type: Localization.Xliff.OM.Core.Segment
Type: Localization.Xliff.OM.Core.Source
Type: Localization.Xliff.OM.Core.PlainText
Content: 'Sentence 2.'

It happens also for <segment> elements.

[RyanKing77]

Thanks for reporting this issue. If you would like to contribute a fix, please do so via a pull request. Otherwise, we will evaluate and prioritize the fix as appropriate.

[RyanKing77]

Upon further examination of the issue, this is by design. If you declare xml:space="preserve" then you will get the expected output. Perhaps you are expecting the default "processing mode" of the OM to be to preserve whitespace? In which case, this is not a bug but a design change.

[RyanKing77]

To clarify a bit more: this is the default behavior of the OM because it is based on .Net XmlReader/Writer which does not preserve by default.

[ysavourel]

Looking more closely at the XML specification, I have to agree: default means the reader does whatever it wants. I had read it as: "The reader can normalize or preserve". But it seems that complete deletion is valid too.

So it's not a bug.

But I would change this as a request for a change in behavior. While remove spaces on outer content is fine, it seems that completely removing spaces in elements that are content like <source> and <target> is probably unwise. I would expect to either normalize or preserve whitespace there.

I'll also post a not to answer your email in the XLIFF list.

[RyanKing77]

Thanks for reporting this issue. If you would like to contribute a fix, please do so via a pull request. Otherwise, we will evaluate and prioritize the design change request appropriately.

[RyanKing77]

Fixed with the following commits https://github.com/Microsoft/XLIFF2-Object-Model/commit/b0dadfe99603fefe599bf5d437cf10f0b03cbe7b https://github.com/Microsoft/XLIFF2-Object-Model/commit/08178a6625b3def5fba7e79378df7e32a8418b63