Language 2026 10 08

The function is a total slugifier: it accepts any string and always returns a slug — lowercase ASCII words joined by single hyphens, never with a leading or trailing hyphen, and `""` when nothing transliterable remains. Accented Latin letters are handled by Unicode NFD decomposition (combining marks are stripped), so `Café` becomes `cafe` without a table entry per accent; letters that do not decompose (`ß`, `æ`, `ø`, `œ`) plus Cyrillic and Greek are covered by an explicit transliteration table, and every run of remaining non-alphanumeric characters — spaces, punctuation, symbols, emoji, CJK — collapses to a single hyphen. It is correct on the inputs the task did not name: empty strings, punctuation-only strings, repeated and edge separators, non-Latin scripts, and very long titles all behave, and a non-string argument raises a `TypeError` that names the type it received instead of returning a misleading empty slug. Two boundaries are deliberate and stated rather than hidden: scripts with no entry in the table (Arabic, Hebrew, Thai, Devanagari) are dropped rather than guess-romanized, and unmapped letters inside the Latin/Cyrillic/Greek ranges are removed so the function never throws on real-world titles.