Louis Ng

Louis Ng said:

I assume Trados is smart enough to leave out the metadata but seems like it's not.

Trados is only as smart as its user... so you need to customse the parser rules accordingly.  XML is very flexible and Trados provide you with the tools to do this.  I suggest you start here:

https://multifarious.filkin.com/2014/06/01/custom-xml/

And if you still need help you will have to share the xml file, or part of it, for anyone to be able to help you.

Louis Ng

Louis Ng said:

I assume Trados is smart enough to leave out the metadata but seems like it's not.

Trados is only as smart as its user... so you need to customse the parser rules accordingly.  XML is very flexible and Trados provide you with the tools to do this.  I suggest you start here:

https://multifarious.filkin.com/2014/06/01/custom-xml/

And if you still need help you will have to share the xml file, or part of it, for anyone to be able to help you.

Hi Paul,

Thanks for sharing the link. It's very helpful but i'm not sure if i can learn how to create filters in such a short time frame. In case i couldn't, is there any other work around? For example, will I be able to tell from the converted Word file which are instructional metadata so that i can leave it in it's original form? or Trados may extract certain wordings in a string of metadata as a segment, which makes it impossible?

I couldn't share the XML on the forum as I've signed NDA.. but if anyone is kind enough to look into this with me, I may can share part of it in a secure way. 

Louis Ng

Louis Ng said:

I couldn't share the XML on the forum as I've signed NDA.. but if anyone is kind enough to look into this with me, I may can share part of it in a secure way. 

All you need to do is pseudo-translate the XML ad save the target.  Once you have done this surely you can anonymise anything left in the header that you think is important to hide and share that?

Louis Ng said:

For example, will I be able to tell from the converted Word file which are instructional metadata so that i can leave it in it's original form?

Frankly this is a ridiculous way to work. If you need to handle XML for a customer then you should learn how to do it properly. It's probably not difficult and if you were to take the time to prepare a sample of the structure we could help you.  We're not interested in the translatable content at all... only the structure of the file.

Hi Paul,

Totally agree. I should just get this over with and try to learn the filters. Below please find the XML with pseudo translation applied so you can have a look at the structure.

XML download link: https://www.dropbox.com/s/4wagoaujdj37ysj/Test%20file.xml?dl=0

I'm able to located the filter settings and understand that I can change the properties to "not translatable", but i'm not sure which filter will fix the problems and whether I need to change the other properties as well. The client listed the issues below and asked me to try changing the parser setting. Hope you can give me a few pointers:

- There are some instructional metadata parts that were translated that should not have been. It should be related to the filter metadata

- Text within graphics was not translated

- Another area of concern that needs to be changed is in the footnote numbers.  Where several footnotes are identified in the same area, the space between the footnote reference numbers are missing. The space was there in the English source doc so somewhere during translation, it’s being removed.  

Many thanks!

Louis Ng

ok - the file is a pretty basic so it should be easy.  My only problem here is that I can't tell what's metadata and what's not because I can't read Japanese... and even if I could now that it's pseudo-translated it makes no sense at all ;-)

What you need to do is open the XML file in a editor and see where the meta data is held?  For example is it held in any of these elements?  It's not obvious from the description and the default XML filetype only extracts text from the elements.  So which of these contains the metadata you wish to exclude?

AutoNumberToken
BoundedText
Cell
Chapter
Classification
content
CopyrightBlock
CopyrightOwner
CopyrightRestrictions
CoverPage
Credits
CustomNote
Definition
DesignData
Distractor
Emph
EndTextWrap
Entry
Figure
FilterMetadata
Footnote
FrontMatter
Glossary
GlossaryItem
Href
IA
ID
InLineVariableText
IntroBlock
Italic
Item
ItemBlock
ItemPara
Lesson
LifeCycle
List
ListPreamble
LOM
MediaObject
Module
Modules
MultipleChoice
name
Notice
Option
OverlayObject
OverlayObjects
OverlayType
ParaBlock
props
QuestionBlock
QuestionStem
Renditions
RichText
Rights
sdoc
sdoc-backup
SimpleBlock
SubList
SubTitle
Table
TableRow
TargetAudience
TargetAudiences
Taxon
TaxonPath
TblBody
TblCol
TblFooter
TblGroup
TblHeader
TblTitle
Term
Title
TitledBlock
Topic
Underline
Version
Web

Once you know this the process is really simple because you'll just make the appropriate elements non-translatable and this will exclude them.

Hi Paul,

Sorry for the late reply as I've been testing the settings with the client's tech team. We have resolved some of the issues but there's one left, in which the text laid on top of graphics is not translated, (Image Overlays?). He suggested that there are several metadata which should not be touched and they asked us to change 4 filters, which we have tried but the problem still exist. Would you mind to help take a look at the files and see if you have any suggestions?

These are the 4 files i'm sharing:

Download link: https://www.sendspace.com/file/qko1uu

1) Source XML

2) Test XML, this was exported after we changed the filter and with pseudo translation applied

3) Image overlays: the client took a screenshot of the metadata which should not be translated

4) Settings change: The filters that the client ask us to change

Many thanks!

Louis

Louis Ng

This isn't particularly helpful. If you provided the settings file it would have been better because I have no idea at all what you have done.  The screenshot of what should not be translated would never be translated by default because they are all attributes so you must have actually created a filetype and told Studio to translate them.

The screenshot of metadata that should not be translated is similarly useless.  BoundedText for example does not contain any text at all, so that rule isn't even doing anything.

Perhaps we can take a quick call because I feel as though I'm wasting my time here when we could probably resolve this quickly if I can speak to you and see what you are trying to achieve?

I wish I can do a quick phone call but I'm actually living in Hong Kong. Plus i'm actually asking for my translator as he is the one who's managing the files. (I was pretending to be the translator earlier as his English level is limited. So he has Trados but i don't and i'm helping him to ask. My apologies). Please bear with me so I can explain it again. If you still don't get it, we can set up a web call to discuss. We really appreciate your help.  

Please find attached the setting file (hope it's right). We have only turned some of the parser rules to "untranslatable" in the attempt to resolve the issues we mentioned. It actually work except for one last problem:

As mentioned previously, my client will import our finished XML file to their design platform (Xyleme) for output, so we applied psuedo translation and sent them a mock XML for trying. They ran it on Xyleme and exported to Word but that the texts in the figures (or table) are not translated. (see attached screenshot, or P.29 of the Word file). They are supposed to be overwritten by pseudo translation if Trados was able to detect it.

I have a feeling that it may have nothing to do with the parser settings. And my translator has only created the project with the default settings. Is there something that we forgot to click so that it can also select the text in Graph/tables for translation?

7318.Files.zip

Louis Ng

ok - we're getting closer.  That table seems to be in here for example:

Note that the translatable text is inside the text attribute of the BoundedText element.  Currently you have excluded this element and have no rule for the attribute:

So I changed it as follows:

In the rule it's this:

This will also pick up this for example... there are more:

Presumably this should also be translated?

Hopefully this will help you understand the difference between elements and attributes at least, so you can probably fix anything else like this.

We tried your suggestion but was not able to locate the text in figure in the exported Word file. Would you mind to check our setting file to see if we have done anything wrong? I have also attached our new XMLand Word for your reference. Many thanks!

New.zip

Louis Ng

You have not applied the rule, you simply made the BoundedText translatable.  I told you to do this:

This is what you did:

You have not extracted the attribute at all.  Here's what you need to do:

Note: I made a small error in describing why you should name the filetype.  You also need to name the File type identifier which you can't change now... you need to do this when you create the filetype in the first place.  If you start to work with XML more you'll be glad you have the ability to do this because it can be very confusing when you can't tell one filetype from another.