Language Server using TextMate grammar for highlighting

[reteps]

Is there a way for a language server for language Y to say "this embedded snippet is language X"

1. without relying on language X having semantic token support
2. without language Y having a textmate grammar
3. Using a pre-existing tree-sitter or textmate grammar for language X

If a language server supports the Semantic Tokens API, but has an embedded language that doesn't support Semantic Tokens, it feels reasonable to allow the language server to use a textmate grammar in order to support syntax highlighting those embedded languages. The alternative is requiring the language server to provide a textmate grammar with support for the embedded language, and then overriding said grammar with the semantic grammar.

It feels like the accepted solution for this situation is request forwarding, but that happens on the Language Client side, and is specific to VSCode.

The other solution I saw was this,

Related:
microsoft/vscode#210475
https://github.com/vsce-toolroom/vscode-textmate-languageservice
https://github.com/microsoft/vscode-anycode/tree/main

Less related:
microsoft/vscode#165788
https://github.com/microsoft/vscode/wiki/Semantic-Highlighting-Overview/72389eecd11607827ec0e345dcd13c89745a3977#token-classification-to-textmate-scopes-list
#1034 (reply in thread)
microsoft/vscode-docker#4183

[RedCMD]

atm the system for implementing embedded/injected languages is rather poor
there isn't really any clean way of doing it; where everything just works
currently you have to manually setup something custom yourself
and often requires both the parent and embedded/injected languages to support the system you're trying to implement
many hack fixes