Romaji to Kana Converter: React & Next.js — Chapter 3, Exercise 1
=====================================================================

TASK
Run convert('kuuki') ("air," くうき) and convert('oneesan') ("older
sister," おねえさん) with real code. Predict each output first, then
verify, and explain any surprise.

SOLUTION

Real, verified output:

  > convert('kuuki')
  { hiragana: 'くうき', katakana: 'クーキ' }

  > convert('oneesan')
  { hiragana: 'おねいさん', katakana: 'オネーサン' }

"kuuki" converts exactly right: "ku" matches として く, then the
second "u" is recognized as a genuine chōon continuation of the "u"
vowel, correctly extending with う in hiragana (the real, correct
extension letter for the u-row) and ー in katakana. Then "ki" matches
normally as き. The hiragana result, くうき, matches the real word
character for character.

"oneesan" is where the real surprise shows up. The katakana output,
オネーサン, is exactly correct -- katakana always uses a plain ー
regardless of which vowel is being extended, so there's no room for
this converter to get it wrong there. The hiragana output, おねいさん,
is genuinely WRONG: the real word is おねえさん, spelled with え, not
い. This converter's default rule extends every e-row long vowel with
い (matching the majority convention, as in せんせい), but "oneesan"
is one of a real, small set of exception words that use え instead --
the exact same kind of unresolvable-from-spelling problem the chapter
already documented for お-row words like とおい and おおきい, just
showing up on the え-row here instead. This isn't a bug to fix; it's
the same honest limitation the chapter already named, independently
confirmed by a different real example.

WHY THIS WORKS AS AN ANSWER
----------------------------
It verifies both outputs with real code rather than assuming they
match the real words, and correctly identifies that "oneesan" isn't
a new bug at all -- it's the chapter's own already-documented
default-rule-vs-exception-word limitation, now demonstrated on the
e-row rather than the o-row.