Support chords in grace notes

[leleogere]

As far as I can tell, it seems that grace notes are simply notes with a zero duration, which does work for simple cases like a single grace note, or a succession of single grace notes. However, grace notes can contain chords, such as in the ASAP score Liszt/Concert_Etude_S145/2:

In that case, I've observed two behaviours:

• 1, 3, 4: Partitura parses the grace notes correctly, however there is no way to tell if the grace notes are forming a chord, or were simply two successive grace notes.
• 2: Partitura fails to parse correctly the grace notes. It links the two grace notes from the second chord (A4-C#5) to the correct main note, and links the two grace notes from the first chord (A4-D5) to the next F# (probably due to some conflicting behaviour between chords and grace note processing):

Code

from pathlib import Path

import partitura as pt
import partitura.score

path = Path("~/Documents/datasets/asap-dataset/Liszt/Concert_Etude_S145/2/xml_score.musicxml").expanduser()

score = pt.load_musicxml(path)
part = score.parts[0]

measure = part.measures[8]
for element in part.iter_all(
    cls=pt.score.Note,
    include_subclasses=True,
    start=measure.start.t,
    end=measure.end.t,
):
    if isinstance(element, pt.score.GraceNote):
        print("   ", element)
    else:
        print(">>>", element)

>>> 576--588 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=C#6
    576--576 GraceNote id=None voice=1 staff=1 type=eighth pitch=F#5 main_note=576--588 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=C#6
    576--576 GraceNote id=None voice=1 staff=1 type=eighth pitch=A5 main_note=576--588 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=C#6
>>> 588--600 Note id=None voice=5 staff=2 type=eighth articulations=(staccato) pitch=A5
    588--588 GraceNote id=None voice=5 staff=2 type=eighth pitch=C#5 main_note=588--600 Note id=None voice=5 staff=2 type=eighth articulations=(staccato) pitch=A5
    588--588 GraceNote id=None voice=5 staff=2 type=eighth pitch=F#5 main_note=588--600 Note id=None voice=5 staff=2 type=eighth articulations=(staccato) pitch=A5
>>> 600--612 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=F#5
    600--600 GraceNote id=None voice=1 staff=1 type=eighth pitch=A4 main_note=600--612 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=F#5
    600--600 GraceNote id=None voice=1 staff=1 type=eighth pitch=C#5 main_note=600--612 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=F#5
>>> 612--624 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=F#5
    612--612 GraceNote id=None voice=1 staff=1 type=eighth pitch=A4 main_note=612--624 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=F#5
    612--612 GraceNote id=None voice=1 staff=1 type=eighth pitch=D5 main_note=612--624 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=F#5
>>> 624--636 Note id=None voice=5 staff=2 type=eighth articulations=(staccato) pitch=A4
    624--624 GraceNote id=None voice=5 staff=2 type=eighth pitch=F#4 main_note=624--636 Note id=None voice=5 staff=2 type=eighth articulations=(staccato) pitch=A4
>>> 636--648 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=B4
    636--636 GraceNote id=None voice=1 staff=1 type=eighth pitch=D4 main_note=636--648 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=B4
    636--636 GraceNote id=None voice=1 staff=1 type=eighth pitch=F#4 main_note=636--648 Note id=None voice=1 staff=1 type=eighth articulations=(staccato) pitch=B4

To summarize, there is two problems with how Partitura handles grace chords:

• There is no way to tell apart two successive grace notes from two simultaneous grace notes in a chord.
  • This is probably not trivial to solve with the current implementation. It would likely be needed to rethink how the Partitura schema works, by defining a Chord object, containing Notes. This would be quite a big change, and I'm not sure that we want to do this. There might be better ways of dealing with that, feel free to propose if you have any idea.
• Successive grace chords seems to result in a bug in the main note attribution, resulting in some grace notes belonging to the wrong main note.
  • This is why I also added the label "bug" to this issue. I could separate them into two issues, but as they are closely linked to the same underlying problem (support of grace chords), I felt like it might be better to let them together. Let me know if you would rather have two issues.