The heuristic mechanism doesn't seem to identify the Vietnamese translations at all

Hi, I have a problem with applying Vietnamese translations for English sources.

  • Summary: Even though I uploaded an existing translation file (*.md) in Vietnamese correctly, the dashboard still shows there is no translated strings (that is, 0% of progress)

  • Steps to reproduce (Download markdown files and screenshots here)

  1. Sources already configured.
  2. Right-click the target file and click Upload Translations.
  3. Select the options and start uploading Vietnamese translation file (file name: creating-accounts.md)
  4. Upload completed.
  5. After reload, there is no change in translation progress.
  6. Download the uploaded translation file.
  7. The download markdown file still shows English strings.

By following the same steps, the heuristic mechanism works correctly most of the cases with Korean translations.

The same problem goes to other Vietnamese translations.

Did I miss something? Is there anything I can try other than “copy-and-pasting” for every string?

Thanks in advance!

Hi @klaytn.docs

Thank you for the detailed explanation of an issue!

The thing is that in the newest version of the md file format the feature to upload the existing translation is available for the following languages (not depending on the language pair combination): Arabic, Chinese Simplified, Chinese Traditional, English, French, German, Italian, Japanese, Korean, Dutch, Polish, Portuguese, Portuguese Brazilian, Spanish, Thai, Turkish, and Russian.

1 Like

Thank you for your answer and information. You mean Vietnamese languages is not supported as for auto-detecting when uploading the existing translation in the latest version the md file format. Right? I have two more questions then.

  1. If the md file format is in any of “older” versions, does it work? (If yes, I’m going to find a way for converting the current md format)
  2. Any alternative way for unsupported languages to get the existing translation recognized? I can’t image how long it will take to copy and paste for tons of strings one by one. :cry:

Hello @klaytn.docs,

I will check with our devs whether there’s any workaround for this and we’ll let you know!

1 Like

Any update on this from your dev? I hope there is a workaround for this by using Crowdin API or CLI, for example. Thanks in advance.

Hello @klaytn.docs,

Thank you for reaching out! Unfortunately, Crowdin API or CLI would not work as a potential workaround now, since they all use the same heuristic mechanism.

We understand the importance of this issue to you and are working to provide a resolution as quickly as possible. We appreciate your patience and will inform you as soon as we have more information.

Let’s keep in touch and have a lovely day :slight_smile:

1 Like

Seems like I’ve found a workaround that works. It’s not the best one, though.

  1. Download a XLIFF file using “Export in XLIFF”.
  2. Replace the English strings inside <target> tags with translated strings and update some attributes such as state and approved.
  3. Upload the new XLIFF file.

I’ll use some automation scripts to accomplish step 2.

Question: Is there any way to download all the translation files in XLIFF, not markdown format? Currently, there seems no other way than downloading each file in XLIFF individually.

Hello @klaytn.docs,

Glad to hear you’ve found a workaround! You can use the XLIFF String Exporter to export your translations in XLIFF format.

As for the direct fix, we’re still working on it. Thank you for your patience and understanding!

Hope this helps :blush:

1 Like

@klaytn.docs, you can also use this application - Translation Package Builder + Crowdin

1 Like

Meanwhile, do you have any progress in this matter? Recently I had to create a new Crowdin project based on the previous project and ran into the same problem when restoring the existing translations.

Hi @klaytn.docs,

As I can see on our side, the task is still in progress

Sorry that it’s taking so long :pray: