Fix #222, #224 — place notes for the clef in force; apply octave=

[sbeitzel]

Fix #222
Fix #224

Summary

Two bugs that had been hiding each other.

• Place notes on the staff for the clef in force, not always treble #222: every note was placed at its treble-clef position, whatever clef the staff carried. The clef glyph and key signature were already right (A clef stated on a K: line never reaches the renderer #126, The baritone clef is drawn on line 4, where a bass clef goes #127, Key signature accidentals are laid out at treble staff positions on every clef #98), so a bass staff showed a correct F clef with treble noteheads under it.
• Apply octave= on K: and V: — it is parsed and never used #224: octave= was parsed on V: and ignored, and K: didn't read it at all.

Both multi-voice conformance fixtures, §7 Zocharti Loch and §14.4 Canzonetta, put their bass voices on clef=bass octave=-2. Treble placement drew bass notes 12 diatonic steps too low, and the missing octave=-2 left them 14 steps too high, so the pages looked nearly right. Each fix alone breaks those fixtures, so they land together.

How

#224 (94b38f3)

• BodyContext tracks the octave= each voice is under. buildNoteEvent and buildChordEvent shift Note.pitch by it, the way abcm2ps's get_note does.
• Scoped as abcm2ps scopes it (get_key, get_voice):
  • On a header K: it applies to every voice.
  • On a body K: or a V: it applies to that voice.
  • It lasts until another octave= replaces it, and octave=0 resets it.
• KeyFieldParser reads octave=.
• Transposition gets an additive statesOctave flag (defaulted, so 2.x API is unchanged). It is what tells octave=0 apart from no octave= at all.

#222 (bfbed66)

• New ClefSpec.staffPosition(of:) counts from the clef's bottom-line pitch. That table was moved out of FloatingVoiceAssigner, which now uses the shared copy.
• The emitter gets the clef in force through configuredForMusic, so notes, chords, stems, ties, slurs, ledger lines and collisions all follow it.
• VerticalLayoutEngine reserves ledger-line space with the same mapping.
• clef=treble-8 and other octave clefs move no notes, as in abcm2ps.
• SharedStaffMerger keeps the treble origin, because it only compares intervals between voices on one staff.

Tests

• New OctaveModifierTests (10 tests):

  • §4.6's example (K:C clef=bass octave=-1 CDEF gives the same pitches as C,D,E,F,). abcm2ps draws the two identically.
  • Chord, grace and tuplet notes.
  • Header, body-K: and V: scope; unstated vs. octave=0.
  • Accidental memory.

• New ClefNotePositionTests:

  • The Place notes on the staff for the clef in force, not always treble #222 reproduction, against abcm2ps -g (8.14).
  • Bottom, middle and top lines of every clef.
  • Octave clefs, and a clef stated on V:.
  • Ledger-space reservation on bass.

  Every non-treble case fails on develop.

• Re-recorded the Zocharti Loch and Canzonetta page snapshots. Both were compared with abcm2ps -g first, and every bass note lands where abcm2ps puts it.

• Full suite passes (793 + 704 + 22).

Not covered

• middle= still doesn't affect placement. The renderer never receives it, and abcm2ps also redraws the clef on a different line for it.
• [K:octave=-1] with no key letter isn't read at all (Accept a K: that names a clef but no key (K:bass, [K:clef=bass]) #223).

🤖 Generated with Claude Code