@tibor.pacalat, can you assign this ticket to @gabirlejenik since he also worked on the related ticket 20054?

Dear @Mazi, I can not assign this to Gabriel, since this is corrupted data from the start, as was in the ticket 20054.

In the LSS file you have attached, you can search for "Muž" (option A2 in question P02). The aid of this option is 183670. It exists only for 4 languages.
Import has nothing to do with the missing options, the options are missing from the original. Please investigate the issue properly before opening the ticket.

If you can provide me with the minimal case to reproduce this issue, then I can confirm and assign it, otherwise I will close this as "won't fix" since there is nothing to fix on already missing data.

@tibor.pacalat: This is NOT the same survey as used before. It was re-created from scratch with the latest Limesurvey 6.x version.

Also, why is the LSS file missing options while exporting the group has all options? Something must be wrong there...

In that case I need original survey (with all the options) to be able to reproduce the issue.
How was the survey recreated?

@tibor.pacalat: The survey was generated by follow the approach outlined under B):
B: Alternative approach:

• Create a blank survey with EN base language
• Import the attached LSG files
• Check the logic file for different languages -> All labels should be there.

Since I can not export that survey (it will have the issues this ticket is about), I will attach all survey groups so you can re-create a working version using the above workaround.
Let me know if there is anything else I can help with.

I have done following:

1. created new survey in EN language
2. added groups
3. added German and French

I can not find an issue, please look into the attached survey.
In the file that you provided (case A) the issue is already visible with those languages, so I assume I don't need to go through all of them.

I have also successfully imported the survey.

Please let me know what you think.

@tibor.pacalat: This is really driving me nuts. I can NOT reproduce at another test system but I CAN reproduce at the customer system.
Can you let me know which PHP version, DB type and version and Limesurvey version (probabla the latest?) you are using?

MySQL, PHP 8.3.20, Limesurvey 6.16.4+260113

Short update: Customer is NOT using MySQL/MariaDB like most others but PostgreSQL, Version number is 12.3_TDE_1.0.

I understood that the labels are already missing on export.
For all questions types or only for some?
Does integrity check find anything weird?

@c_schmitz, I have exported the affected survey directly from the customer system running LS 6.15.22+251103 on PostgreSQL (libpq) Version 13.23.

Analyzing the XML structure reveals some interesting details:
limesurvey_group_10104-edge.lsg: The group export contains 54 languages, 57 language-independent answer options, and 3,078 directly counted localization records. No duplicates or missing language assignments were found.

limesurvey_survey_666633-edge.lss: For the same group (gid 10104), the full survey export also contains 54 languages and 57 language-independent answer options, but 3,184 localization records were counted. These include 492 exact duplicates, leaving only 2,692 unique records. In total, 386 expected answer-language assignments are missing; answer IDs 183671 and 183676 have no localization records at all.

This indicates that the full survey export produces an inconsistent <answer_l10ns> section, while the standalone group export is complete and consistent.

Please check if the attached file fixes the issue.