Samuel Baldus

I'm not sure... works fr me:

I guess we're missing the appropriate information to replicate your problem.  Proper sample file, maybe your filetype settings?

Hi Paul Filkin

Odd.

I have attached a really basic XLF, and XlfToXml file type.

Maybe you can see where I am going wrong?

Thanks

Sam

XlfToXml.zip

Samuel Baldus

I believe this may be related to the namespace.  I added this to your settings:

//p0:note[not(contains(text(),'Do not catch this'))]

Then added the uri when prompted:

urn:oasis:names:tc:xliff:document:1.2

Now I see this and it looks as though it's working nicely:

Samuel Baldus

I believe this may be related to the namespace.  I added this to your settings:

//p0:note[not(contains(text(),'Do not catch this'))]

Then added the uri when prompted:

urn:oasis:names:tc:xliff:document:1.2

Now I see this and it looks as though it's working nicely:

Thanks Paul, that seems to work well as a translatable item.

It doesn't seem to work though, if you want to use it as a Comment XPath:

    
<?xml version="1.0" encoding="utf-8"?>
<xliff version="1.2" xmlns="urn:oasis:names:tc:xliff:document:1.2">
 <trans-unit>
 <target>First source term</target>
 <note>Catch this</note>
 </trans-unit>
 <trans-unit>
 <target>Second source term</target>
 <note>Do not catch this</note>
 </trans-unit>
</xliff>

 

Is that at all possible?

Thanks

Samuel

Samuel Baldus

Apologies for the delayed response.  I tried it in a quick few minutes yesterday and could not make it work.  So, I played around a bit today after thinking about it on a long walk ;-) I can only make this work if I remove the namespace parameters and use local-name instead.  So these:

Then these rules:

//*[local-name()='target']

../*[local-name()='note'][not(contains(text(),'Do not catch this'))]

I think Patrik Mazanek may be able to comment on this.  To me it looks as though the comment XPath doesn't respect the use of namespaces, but I could easily be doing something wrong!

But hopefully this gives you a way to handle the file as you wish.

Hi Paul Filkin

Thanks for your efforts, nothing like a good walk to clear the mind, I tried your solution and it seems to work.

The only problem I am facing now is that this method does not work when the HTML Embedded Content Processor is used.

Kind regards

Samuel Baldus

Samuel Baldus said:

The only problem I am facing now is that this method does not work when the HTML Embedded Content Processor is used.

Hmnn... nothing like being drip fed the requirements.  Interesting how this thread has evolved!

This indeed will be a problem.  We can across this when we did the MultilingualXML filetype and actually built our own solution to support embedded content in XML in a better way wrt segmentation and being able to benefit from both using an embedded content processor (like html) and also use regex to catch anything the html processor will miss.

During the development of that filetype we also discussed comments and omitted it for the time being while we waited to see if anyone really wanted that.  I think you are the first person to present a usecase where it would be helpful.  And your usecase is an interesting one!  So we'll take a look at this for a future release... when we get some time as we're ridiculously swamped at the moment.  For now I think you are probably stuck.

Although as before... providing more details on your requirement and even your filetype settings would be more helpful than drip-feeding the information.  Maybe there will be a solution after all?

Hi Paul Filkin

Apologies for not giving you everything in one go.

My approach when something isn't working as expected, is to strip it down to be bare minimum and go from there.

If the base level functionality does not work or make logical sense, then it's easier to trouble-shoot that rather than getting lost in a haystack of potential issues.

Thanks for the explanation though, I hope to see the comments in embedded content in future releases.

Samuel Baldus

Samuel Baldus said:

I hope to see the comments in embedded content in future releases.

I think this is something you should probably add to the ideas site as well along with an explanation of why you would need this?  It could be quite tricky to achieve and if you're the only person who wants it I doubt it'll get a lot of traction.  When we looked at it our thoughts were just about exposing the possibility to extract a comment using XPath.  I don't think we even dreamed about this also containing tagged content we would need to process further!

Hi Paul Filkin

I think there might be some crossed wires here.

The comments themselves do not contain embedded content.

The Comments Xpath does not work when embedded content is enabled in the XML file type. 

Samuel Baldus

Thank you for that clarification.  I was thinking something different.  I just tested this and you're right.  I'm going to raise both problems discussed in here with support so they can validate, explain better, and raise with development.

Thanks Paul Filkin

Samuel Baldus

I have an update on this in two parts.

1. the fact you have to use local-name and can't use the namespace prefix is a confirmed bug.  I'll have a bug number for this in due course... hopefully next week.
2. the use of embedded content problem is only a problem because I didn't now how to work around it.  According to support this is the intended behaviour because when using embedded content the text moves from the xml filter to the html one where there is no option for comments.

There is a way to deal with this though by using the feature to tackle it through a defined document structure.  So you add structure like this for example:

Then use this option for the embedded content:

Unfortunately this means you can't use the embedded content filetypes such as HTML and will have to create the rules you need.  Not a bug, but definitely worth raising as an enhancement idea.

It also means you get a pretty horrible experience if there is embedded content on the comments:

But it does work.

A long old thread and quite a lot of useful information in here I think.