Thread: Translating AutoCAD files MemSource vs. SDL/Trados

I've been running some tests comparing these two CAT tools and would like to open this thread to share with other's experiences.

As per my previous posts, I use TranslateCAD to export the text. It does so by creating to TXT file with some sort of "tokenization" of the file with a mixup of translatable content and code. From each pair of TXT files, you need to work with the one named "trans1" that bears the translatable text.

Right from the beginning, even though I've found that (at least for me) it only worked when working with DXF (AutoCAD) files saved as 2007-2009 ASCHII, the TXT encoding is apparently automatically switched into Unicode when you work in the SDL environment. This does not happen when working with MemSource, where I need to open the completed TXT file and resave it in Unicode before being able to reimport it back into the AutoCAD file.

Hi Gentle!
How you doing?

Yes, I quite remember your recent posts regarding AutoCAD files, here.
Well, I haven´t experienced AutoCAD files in Memsource yet. I did work with TranslateCAD with Trados, and I remember the process was quite smooth. You get both txt, then you translate txt1 and then just import it back to TranslateCAD...
From what you experienced, it seems that Trados does that Unicode switch internally and that´s why we never knew about it?

So, when you work in Memsource, once you get the complete version, which encoding the txt file has?

Originally Posted by analaura

So, when you work in Memsource, once you get the complete version, which encoding the txt file has?

Ana, it comes out as an ANSI file. I'm not sure if you can set something up when you upload the file so that it can be managed as a Unicode one. In any case, all you need to do is download the "completed" file, open it in Notepad (or any TXT editor) and save it, but not before selecting Unicode as the final encoding (from a drop-down at the Windows "Save" pop-up window). Then you can import it back with TranslateCAD as usual.

Hi Gentle

it sounds pretty straight forward, I mean, It would be much better if we could avoid all that Unicode saving, but it´s not big deal compared to the huge advantages of the Translate CAD... What do you think?
I used to extract the text from the autocad files and paste text into MS Excel sheets and then do all the backward work onto the autocad files, a very costly and time consuming option...

Originally Posted by analaura

I used to extract the text from the autocad files and paste text into MS Excel sheets and then do all the backward work onto the autocad files, a very costly and time consuming option...

That's definitively an option. And here comes one more issue when using MemSource straight with the TXT files: it creates an empty segment that needs no translation and cannot even be confirmed. That is, it doesn't even count as a segment! In other words, it remains empty and unconfirmed even though the file will look as 100% completed in the portal. It does not interfere with the processing, but it's weird and a bit annoying.

Hi gentle,

Yes, it's true. It creates empty segments in the target. The only thing is that the QA identifies them as "empty target" so you need to leave that option unmarked.

mm, I didn´t know about that... yes, it´s a bit annoying, since you have to change status to Complete manually. Hopefully, in the new Memsource realeases, we might enojoy a more smooth process for AutoCAD files...

I'm not sure why, when you pull out the "completed" TXT "trans-1" file from the cloud, the encoding varies from file to file (sometimes ANSI, others UTF-8 or even Unicode). The fact is, at least for my version of TranslateCAD, all files can only be reimported (reconstructed) if they are in Unicode...

I've noticed that dealing with source files in German (TXT extracted with TranslateCAD), two different issues were found: complex characters like vowels with diaeresis or the particular ß used in German are corrupted and replaced by weird symbols. It's annoying as it infects the translation memory with unwanted characters and can also interfere with leverage.

I've jumped one more hurdle: When pulled out of MemSource, some of the TXT files get corrupted in the sense that paragraph marks are erased. The original source layout (line by line) turns into one single paragraph. Thank God this is easily solved by selecting all text, copy/pasting it in an Excel spreadsheet and doing the reverse back into TXT and saving the file one last time. However, it's quite annoying! One more reason to advocate SDL's Studio to process AutoCADs.