Add the Ukrainian conjunction й (#267 comment, #269 follow-up)

[derek73]

Where this came from

A reply on #267 noted, while confirming that a bare Cyrillic letter should read as the conjunction rather than an initial:

In Russian, it is и and in Ukrainian, it is either і or й.

We ship и, і and та — not й. #269's candidate list read "Cyrillic conjunctions: и; Ukrainian і/та", so й was never considered and rejected; it just never made the list.

й is the euphonic alternate of і. Ukrainian picks between them by the surrounding vowel and consonant, not by meaning — "Олесь і Олена" but "Марія й Петро" — so real data carries both spellings and recognizing one of the two mis-parses the other:

'Олесь і Олена Коваленки'  ->  given: 'Олесь і Олена'   family: 'Коваленки'          ✓
'Олесь й Олена Коваленки'  ->  given: 'Олесь'  middle: 'й Олена'  family: 'Коваленки'  ✗

The change

One entry in nameparser/config/conjunctions.py. Pure data — no code path changes — and it inherits both guards the existing single-letter conjunctions rely on:

• The single-letter carve-out (Google Code issue 11, the "john e smith" bug) still holds it back until the segment has enough rootname pieces, so 3-piece Іван й Олена stays split exactly as і and y do.
• The dotted branch of the initial regex is Unicode-aware, so a punctuated Й. remains an initial and is untouched.

That second one seemed worth a standing test rather than a comment: и/і/й each open real given names (Игорь, Ірина, Йосип), so the shipped entries would be a trap if membership could outrank a punctuated initial. test_267_dotted_cyrillic_initial_wins_over_conjunction pins all three dotted forms.

Tests

Two rows added to the #269 non-Latin parametrization — the title-chain shape the и/і/та rows use, and the given-name join from the issue comment — plus the dotted-initial guard. Both new rows were watched failing on the old data before the entry landed.

2699 passed, 32 skipped, 11 xfailed; mypy and ruff clean.

Notes

• Differential harness needs nothing. No name in any corpus*.jsonl contains й, so there is no new diff; and had one appeared, the existing feat(#269) non-Latin titles/conjunctions recognized rule in expected_changes.toml already classifies exactly this shape (Cyrillic, fields title/first/middle).
• Release-log entry added under 2.1.0 → Behavior Changes.

Also checked, no action needed

The rest of that comment holds up against current behavior:

• И. Иванов, І. Франко, И. С. Петров all parse as initials — the typography point Should a bare 'И' parse as an initial or the conjunction 'and'? (e.g. 'Хосе И Мария Сантос') #267 already settled.
• А. П. и Е. Н. Ивановы gives first: 'А.', middle: 'П. и Е. Н.', last: 'Ивановы'. Lossy-looking, but identical to the Latin analogues A. P. and E. N. Ivanovs and A. P. y E. N. Santos — the general multi-person-string limitation, not a Cyrillic bug.
• Plural family names (Ивановы → Иванов/Иванова) are morphological normalization; nothing analogous exists for any language here.
• Cyrillic conjunction coverage is otherwise complete after this: и serves ru/bg/sr/mk, і/й/та serve uk/be.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.45%. Comparing base (d09ac7f) to head (60943ed).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #306   +/-   ##
=======================================
  Coverage   98.45%   98.45%           
=======================================
  Files          41       41           
  Lines        2776     2776           
=======================================
  Hits         2733     2733           
  Misses         43       43           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.