Bernadette Haury 

There are certainly ways forward with this.

The approach I would take is:

1. Export TM's to TMX
2. Create EN > FR, DE project(s), using your TMX files as the native file type
3. From your SDLXLIFF file(s), use the export to Excel app
4. Merge the Excels so you have complete English column followed by 2 dedicated columns for FR and DE 
   Example A = English, B = FR and C= DE
5. Create new project using your merged excel file, with the Multilingual Excel file type
   

When working your way through your newly created project based on your merged Excel, you will now find the "gaps" between your FR and DE translations.
You can either create new TM's or readd your existing TM's for update

There is also a feature called AnyTM that offers some flexibility when working with TM's that dont match language pairs. Details of which is here

I hope this helps and keen to see if anyone has other ideas

Lyds

 Bernadette Haury 

There are certainly ways forward with this.

The approach I would take is:

1. Export TM's to TMX
2. Create EN > FR, DE project(s), using your TMX files as the native file type
3. From your SDLXLIFF file(s), use the export to Excel app
4. Merge the Excels so you have complete English column followed by 2 dedicated columns for FR and DE 
   Example A = English, B = FR and C= DE
5. Create new project using your merged excel file, with the Multilingual Excel file type
   

When working your way through your newly created project based on your merged Excel, you will now find the "gaps" between your FR and DE translations.
You can either create new TM's or readd your existing TM's for update

There is also a feature called AnyTM that offers some flexibility when working with TM's that dont match language pairs. Details of which is here

I hope this helps and keen to see if anyone has other ideas

Lyds

Unknown said:

I hope this helps and keen to see if anyone has other ideas

Indeed a smart solution.  But given the use of AI these days I thought I'd try a different approach for fun.  It didn't take very long and you may be interested.

Bernadette Haury 

Bernadette Haury said:

Does anyone know if that's possible anyhow?

Here's a way using the concept of multilingual TMX files.  I created a Python script with the help of ChatGPT that can take the English to French TMX, and the French to German TMX and merge them to create a multilingual TMX with all three languages in there.  Then I can import that TMX into a French to German SDLTM and it will populate with English to German.

Here's the script: 

import xml.etree.ElementTree as ET

def merge_tmx(eng_fr_file, fr_de_file, output_file):
    # Define the namespaces (if any other namespaces are used, add them here)
    namespaces = {
        'xml': 'http://www.w3.org/XML/1998/namespace',
    }
    
    # Register the namespace
    ET.register_namespace('xml', namespaces['xml'])
    
    # Parse the English-French TMX file
    tree_eng_fr = ET.parse(eng_fr_file)
    root_eng_fr = tree_eng_fr.getroot()

    # Parse the French-German TMX file
    tree_fr_de = ET.parse(fr_de_file)
    root_fr_de = tree_fr_de.getroot()

    # Create a dictionary to hold French to German translations
    fr_de_dict = {}
    for tu in root_fr_de.find('body'):
        french_seg = tu.find(f"tuv[@xml:lang='fr-FR']", namespaces).find('seg').text
        german_seg = tu.find(f"tuv[@xml:lang='de-DE']", namespaces).find('seg').text
        fr_de_dict[french_seg] = german_seg

    # Iterate through the English-French TMX and add German where the French matches
    for tu in root_eng_fr.find('body'):
        french_seg = tu.find(f"tuv[@xml:lang='fr-FR']", namespaces).find('seg').text
        if french_seg in fr_de_dict:
            # If the French segment matches, add the corresponding German segment
            german_tuv = ET.Element('tuv')
            german_tuv.set(f"{{{namespaces['xml']}}}lang", 'de-DE')  # Corrected way to add namespace
            german_seg = ET.SubElement(german_tuv, 'seg')
            german_seg.text = fr_de_dict[french_seg]
            tu.append(german_tuv)

    # Write the merged TMX to a new file
    tree_eng_fr.write(output_file, encoding='utf-8', xml_declaration=True)

# Prompt for file names
eng_fr_file = input("Enter the path of the English-French TMX file: ")
fr_de_file = input("Enter the path of the French-German TMX file: ")
output_file = "en-GB_fr-FR_de-DE.tmx"

merge_tmx(eng_fr_file, fr_de_file, output_file)
print(f"Multilingual TMX file created: {output_file}")

And here's a video explaining how to use it:

Many thanks for your reply. I gave it a first try, but the export to Excel took very long, as it's a large TM, and I had to cancel it as I could not continue working while the export was working. I might give it another try over night.

Thanks!

Many thanks for your reply as well. I'll definitely give it a try, thanks.

Bernadette Haury 

Bernadette Haury said:

I'll definitely give it a try

You should... this way you will also retain the document structure for better context.  And it's fast.

Paul Thanks a lot for this, I tried and it was indeed very fast. The aligned TM is quite large, but at a first glance, the results look quite good. Thanks for trying this new technology :)