CHatGPT to translate ancient Articulate Xliff files in Rise

Hello all, 

First and foremost, Articulate Rise's xliff files aren't compatible with DeepL. That's really really, really, blocking us to do quick learning courses and to deliver them worldwide to our 10.000+ company's learners.

Synthesia offers translation of courses within its services, no export needed. Now that Articulate has AI, maybe machine translation will be next?

While waiting for this, here's how we managed to get Articulate xliff files translated by ChatGPT ... without the recommanded tools that we don't have a licence of (Smartcat, ...).

0. Have Notepad++ or a good alternative installed for code manipulations. Xliff is a kind of xml markup language.

1. Duplicate the course and download the course_translation.xlf file and open it in Notepad++. 
2. Use Notepad++ to separate chunks 7000 chars of code in between transunit tags </trans-unit>CUT_HERE<trans-unit id=...">
3. Open ChatGPT, and give him first a context prompt : "Hello ChatGPT, I would like to have you translate xml code for me in {target language}. Would it be possible for you to just translate the text in between the xml tags, but to keep all the code exactly as is, without changing it, eg. without touching, indenting, or reformatting the tag structure? Could you also keep the vertical bar character, "|", as is, and please don't convert &lt;p&gt; into <p> and &lt;/p&gt; into </p>. Please confirm the instructions."
4. Either paste the code or if it doesn't keep the xml tag structure while pasting, use a buffer "temp.xml" file where you paste your code chunks and then upload this file. Of course if you have a chatGPT licence, this gets easier ;)
5. Copy/paste the resulting code to replace your code chunks with the translated one. 
6. In Notepad++, search for "source>" and replace with "target>" to tell Rise to load the translations, and save your file.
7. Load the resulting translated_course.xlf in Rise
8. Enjoy a coffee and to have spared yourself some 8+ hours of tedious work.

This process might need some tweaking. You might need to upload the translated_course.xlf step by step if not all the course in translated in one go : if Rise detects any problem, it stops translating. 

Especially, ChatGPT translates the "|" into "\n" and "&lt;p&gt;" into "<p>" and that will make fail to load the translation. Check for any "\" (antislash) character left.

... Once you have some working process, doing other languages / modules is quite fast :)

Hope it helps, hope Articulate adds translation into Rise soon! Cheers.