TC String Encoding/Decoding Defaults Advanced Language Codes to English (EN)

[pavankumarupadhyayula]

Description

There is a critical issue with the @iabtechlabtcf/core library where advanced language codes are not properly preserved during the TC string encoding and decoding process.

Problem Statement

When advanced language codes such as PT-BR (Portuguese - Brazil) or SR-LATN (Serbian - Latin script) are used, the implementation incorrectly defaults to English (EN) instead of preserving the specified language code as required by the IAB TCF specification.

Affected Component

• File: ConsentLanguages.ts
• Location: modules/core/src/model/ConsentLanguages.ts in the iabtcf-es repository
• Repository: https://github.com/InteractiveAdvertisingBureau/iabtcf-es

Current Behavior

1. SR-LATN has been added to the supported language Set in ConsentLanguages.ts
2. During the encoding process, when advanced language codes like SR-LATN or PT-BR are specified, the implementation defaults to English (EN)
3. When the TC string is decoded, the language shows as EN instead of the originally specified advanced language code (SR-LATN, PT-BR, etc.)

Expected Behavior

The full language code (e.g., SR-LATN, PT-BR) should be preserved throughout the encoding and decoding process, ensuring that the exact language specification is maintained in the TC string.

Impact

This issue affects:

• Consent strings generated for regions requiring specific language variants (e.g., Brazilian Portuguese, Serbian Latin script)
• Decoding accuracy when reading consent strings
• Compliance with IAB TCF specifications for language handling

Steps to Reproduce

1. Set the consent language to an advanced language code (e.g., SR-LATN or PT-BR)
2. Generate/encode the TC string
3. Decode the TC string
4. Observe that the decoded language shows as EN (English) instead of the specified advanced language code (SR-LATN or PT-BR)

[HeinzBaumann]

@pavankumarupadhyayula You raise this against the gpp-es but your talk about the standalone TCF es library. Am I correct that the bug you report is in the iabtcf-es library?

[HeinzBaumann]

The reason this happens is because the TCF specification only support ISO-639-1 which is a two language letter code. The specification allows 12 bits for the language code (6 bits per letter). Something like SR-LATN is correctly encoded as SR and decoded as SR. In the iabtcf JS library it will return SR-LATN if SR is found. This seems to be working correctly.

[HeinzBaumann]

I have not check yet the iabgpp-es library.