Compatibility Decompostion may not work as expected

[ManyTheFish]

Check if:

• "Ａｐｐｌｅ" is converted to "Apple"
• "ｱｯﾌﾟﾙ" is converted to "アップル"

when normalizing them, the CompatibilityDecomposition normalizer should do the job.

[sahilmadaan048]

Verified: CompatibilityDecompositionNormalizer handles these cases correctly.

"Ａｐｐｌｅ" → "Apple"
This full-width Latin string is correctly normalized to its ASCII form using NFKD, as expected.

"ｱｯﾌﾟﾙ" → "アップル"
This half-width Katakana string is normalized to its full-width equivalent (using decomposed characters, e.g., プ = フ + ゚), which visually and semantically corresponds to "アップル".

The current implementation of CompatibilityDecompositionNormalizer leverages char.nfkd(), which performs compatibility decomposition in line with Unicode TR15. Both cases are handled as expected under this normalization.

Conclusion: No changes are needed; the normalizer already supports the expected conversions.

[ManyTheFish]

Hey @sahilmadaan048, thank you for your investigations, 👍

closing this issue!