Stefano,Decomposition is NOT always safe or appropriate when converting from script-to-Roman. For example, Cyrillic has many characters with diacritical marks that should not be encoded as separate characters, in particular the Йй Ёё. In Latin transliteration, these diacritical marks are encoded separately.Randy Barry - Рэнди Барри - ᠷᠧᠨᠳᠢ ***@***.*** Mar 16, 2025, at 10:17, Stefano Cossu ***@***.***> wrote:This PR adds a normalization step to both S2R and R2S. It converts all pre-composed characters in the source into their decomposed form (combining diacritic + base symbol). With this step, conversion tables only need to address tokens in the decomposed form. If this is the intended behavior for Scriptshifter, please approve. You can view, comment on, or merge this pull request online at:   #189 Commit Summary aefb899 Normalize precomposed Unicode characters. 47df258 Fix typo. File Changes (1 file) M scriptshifter/trans.py (10) Patch Links: https://github.com/lcnetdev/scriptshifter/pull/189.patch https://github.com/lcnetdev/scriptshifter/pull/189.diff —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because your review was requested.Message ID: ***@***.***>

Yes, I think normalizing precomposed to decomposed for Latin script should be safe. Since the old MARC environment is very specific, it may be best to control exactly which Latin Precomposed characters get decomposed.Randy Barry - Рэнди Барри - ᠷᠧᠨᠳᠢ ***@***.*** Mar 16, 2025, at 19:39, Stefano Cossu ***@***.***> wrote: Would it be safe to apply to R2S only?—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because your review was requested.Message ID: ***@***.***> scossu left a comment (lcnetdev/scriptshifter#189) Would it be safe to apply to R2S only? —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because your review was requested.Message ID: ***@***.***>