add-data-source fails with "Failed to sanitize string" for tables whose display name is fully non-Latin

[Mosh-K]

Bug

Running:

npx power-apps add-data-source --api-id dataverse --resource-name <table-logical-name> --org-url <org-url>

fails with:

Failed to update database references: Failed to sanitize string <non-Latin-display-name>

for any table whose display name has no ASCII letters/digits (e.g. Hebrew, CJK, Arabic). Generation aborts; no table can be added. (pac code add-data-source fails identically — it has its own copy of the same logic.)

Cause

In @microsoft/power-apps-actions/dist/CodeGen/shared/nameUtils.js, sanitizeName() runs name.replace(/[^a-zA-Z0-9_$]/g, '_') on the table display name and then throws when the result is all underscores. A fully non-Latin display name collapses to ______ → throw. The logical and entity-set names are ASCII and sanitize fine; only the display name triggers it.

Fix

Preserve Unicode letters/digits in the replacer:

name.replace(/[^\p{L}\p{N}_$]/gu, '_')

Related theme to #329 but a distinct failure mode (that's label corruption; this is a hard crash) — not a duplicate.

[eschavez]

Thanks for the detailed report and the precise root cause analysis, @Mosh-K!

We've confirmed this: sanitizeName() in our codegen uses /[^a-zA-Z0-9_$]/g which collapses a fully non-Latin display name to all underscores, then throws. The logical and entity-set names are ASCII and sail through fine — only the display name hits this path for shared-name connector schemas.

Workaround (while we ship the fix): If your environment allows it, you can temporarily rename the table display name to include at least one ASCII letter/digit in your Dataverse org settings, run pac code add-data-source, then rename it back. We recognize that's not always practical for production tables.

We're working on a proper fix (Unicode-aware regex so Hebrew, Arabic, CJK and other non-Latin display names are preserved in generated identifiers). We'll update this issue when the fix ships.