Extracted message not namespaced when using next-intl

[oscar-b]

When extracting a string in a project with next-intl, it isn't saved to the file with the namespace nesting which is defined in the source file:

const t = await getTranslations('RootLayout');
t("new-message")

Instead of ending up in as:

{
  "RootLayout": {
    "new-message": "123"
  }
}

it ends up as:

{
  "new-message": "123"
}

Is there any special configuration that's needed to get this behavior?

After extracting, the plugin reports:

ERROR: 'RootLayout.new-message' not found in source with language tag 'en'

If I manually move the new message to the correct namespace in the JSON, the plugin detects this properly and displays it as it should, it's just the extraction that doesn't work.

[samuelstroschein]

Yeah tricky. Sherlock has no concept of a namespace.

The implementation of matching a message like RootLayout.new-message happens via the next-intl plugin. Extending the next-intl plugin to report a namespace for a given file as well as adding an onExtract callback to Sherlock could be a solution.

In the meantime, either not using namespaces or manually fixing the key is the way to go. Sherlock has a feature freeze till lix 1.0 and the inlang sdk 2.0 are out, see https://opral.substack.com/p/focus-shift-from-inlang-to-lix

[oscar-b]

@samuelstroschein All right, was mostly confused since parts of the plugin handled namespaced strings and the other part didn't :)