symbols.txt

First a glossary of terms:
	AnnNote is a musicdiff-annotated note/rest/chord (by default, there are no
	    chords, only the notes within them)
	AnnVoice is a musicdiff-annotated voice (by default, there are no AnnVoices,
	    only the AnnNotes within them)
	AnnLyric is a musicdiff-annotated lyric syllable
	AnnExtra is a musicdiff-annotated musical object such as dynamic, text, clef,
	    keysig, timesig, metronome mark, etc
	AnnMeasure is a musicdiff-annotated measure
	AnnPart is a musicdiff-annotated staff
	AnnStaffGroup is a musicdiff-annotated staff group (as denoted by
	    braces/brackets/etc to the left of the score)
	AnnMetadataItem is a musicdiff-annotated metadata item, such as composer,
	    date of composition, etc
	AnnScore is a musicdiff-annotated score

AnnNote symbol count (rest, note, chord):
    (If Voicing is not specified explicitly on the command line, there are no
        chords, only the individual notes from the chords)

    For each note (1 rest, 1 note, or N notes in a chord):
        1 symbol for the pitch (even if unpitched)
        1 symbol if there is an accidental (even a triple flat is just 1 symbol)
        1 symbol if a tie starts on the note (we don't count tie stops)

    1 symbol for note head type: quarter, half, whole, breve etc
    1 symbol for each dot (e.g. double-dotted quarter note gets 2 symbols for dots)
    1 symbol for each flag or beam
    1 symbol for each enclosing tuplet
    1 symbol for each enclosing tuplet's style (is it visible? bracketed?)
    # (note head type + dots + flags/beams + tuplets implies visual duration)
    1 symbol for each articulation (staccato, tenuto, etc)
    1 symbol for each expression (mordent, trill, etc) (in future add 1 for
        ornament accidental?)
    1 symbol if a grace note
        1 symbol if that grace note has a slash
    1 symbol if there is abnormal note head shape (diamond, etc)
    1 symbol if there is abnormal note head fill
    1 symbol if there are parentheses around note head
    1 symbol if stem direction is specified
    1 symbol if there is a space (hidden rest) before the note
    1 symbol for all specified style info (e.g. color, size, etc) (in future
        might compare individual style entries)

AnnNote diff symbolic edit count:
    for each note (1 rest, 1 note, or N notes in a chord):
        pitch (line/space position) diff: If Voicing is not explicitly specified
            on the command line, we only compare notes that have the
            same line/space (vertical position) and offset (horizontal position).
            Any notes with no same-position note to compare against are deleted
            or added (with symbolic edit cost == note symbol count). If Voicing
            _is_ explicitly specified on the command line, a pitch diff is 2
            symbols (delete then add).
        accidental diff: 1 per accidental added or deleted
        tie start diff: 1 (add or delete the tie)
    note head type (quarter/half/whole/breve/etc) diff: 2 (delete then add)
    dots diff: 1 for each dot added or deleted
    grace type diff (acc vs unacc vs not grace): 2 (delete then add)
    grace slash diff: 1
    beams/flags diff: Levenshtein distance between lists of beams/flags
    tuplet diff: 1 for difference in tuplet actual count/normal count + 1 for difference in
        visibility, bracketing, etc.  If there is more than one (nested) tuplet, instead of
        1 + 1, use Levenshtein distance between lists of tuplets (actual/normal) + Levenshtein
        distance between lists of tuplets (visibility/bracket).
    articulation diff: Levenshtein distance between lists of articulations
    expression diff: Levenshtein distance between lists of expressions
    note shape diff (diamond, etc): 2 (delete then add)
    note head fill diff (beyond the usual quarter vs half): 1 (delete then add)
    note head parentheses diff: 1 (delete or add)
    "space before" diff: 1 (representing a horizontal shift of the note)
    note stem direction diff: 1 if adding or deleting stem, 2 if changing stem
        direction
    note style diff (e.g. color, size, etc): 1

-------------------------------------------------------------------

AnnExtras have custom counts by type, as follows:

clef symbol count:
    1 symbol for clef itself (e.g. G2, F4, C3, G2-8, etc)
    1 symbol if styled in any way

clef diff symbolic edit count:
    clef diff: 1
    clef style diff: 1

timesig symbol count:
    1 symbol for each number or symbol (e.g. 3/8 is two symbols, common time (C)
        is one symbol)
    1 symbol if styled in any way

timesig diff symbolic edit count:
    timesig diff: 1 per delete and/or add of number or symbol (3/8 vs 6/8 is a
        delete of the 3 and an add of the 6, so two symbols; 4/4 vs C is a
        delete of both 4s and an add of the C, so three symbols)
    timesig style diff: 1

keysig symbol count:
    1 symbol for each accidental in signature (exception: no sharps/flats gets
        1 symbol)
    1 symbol if styled in any way

keysig diff symbolic edit count:
    keysig diff: 1 per delete and or add of accidental (two flats vs no
        flats/sharps is two symbols deleted; 3 sharps vs 4 flats is 3
        symbols deleted + 4 symbols added; 5 flats vs 4 flats is one symbol
        deleted)
    keysig style diff: 1

tempo symbol count:
    1 symbol for anything like "<note> = 80"
    len(string) symbols for any other string (like "Andante")
        - both may be present (e.g. "Andante <note> = 80" is len(“Andante”)+1=8)
    1 symbol if styled in any way

tempo diff symbolic edit count:
    diff in "<note> = 80" portion: 2 (delete then add)
    diff in other string portion: Levenshtein distance between other string portion
    tempo style diff: 1

barline symbol count:
    1 symbol for non-regular barline (e.g. double, light-heavy, repeat, etc)
    1 symbol for repeat direction, if present
    1 symbol for repeat count, if specified
    1 symbol for barline fermata if present
    1 symbol for non-normal barline fermata shape (if barline fermata present)

barline diff symbolic edit count:
    barline type diff: 2 (delete then add)
    repeat direction diff: 2 (delete then add)
    repeat count diff: 2 (delete then add)
    barline fermata diff: 1 if delete or add, 2 if change type of fermata
        (inverted or not)
    barline fermata shape diff: 1 if delete or add non-normal shape, 2 if
        change non-normal shape

ottava symbol count:
    1 symbol for ottava type (e.g. 8va, 8ba, 15ma)
    1 symbol for ottava duration
    1 symbol if styled in any way

ottava diff symbolic edit count:
    ottava type diff: 2 (delete then add)
    ottava duration diff: 1 (horizontal shift of ottava end)
    style diff: 1

direction symbol count:
    len(string) symbols for direction string
    1 symbol if styled in any way (italics, bold, color, placement, etc)

direction diff symbolic edit count:
    string diff: Levenshtein distance between strings
    style diff: 1

dynamic (text, not wedge) symbol count:
    1 symbol for the name (e.g. fff)
    1 symbol if styled in any way (italics, bold, color, placement, etc)

dynamic diff symbolic edit count:
    string diff: 2 (delete then add)
    style diff: 1

dynamic wedge symbol count:
    1 symbol for the wedge type (crescendo or diminuendo)
    1 symbol for the wedge duration
    1 symbol if styled in any way (placement, etc)

dynamic wedge diff symbolic edit count:
    wedge type diff: 2 (delete then add)
    duration diff: 1 (horizontal shift of wedge end)
    style diff: 1

slur symbol count:
    1 symbol for the slur duration
    1 symbol if styled in any way (placement, etc)

slur diff symbolic edit count:
    duration diff: 1 (horizontal shift of slur end)
    style diff: 1

arpeggio symbol count:
    1 symbol for arpeggio type (up, down, undirected, non-arpeggiated)
    1 symbol if span length > 1 (e.g. if arpeggio spans multiple chords)
    1 symbol if styled in any way

arpeggio diff symbolic edit count:
    arpeggio type diff: 2 (delete then add)
    span length diff: 1 if old or new span length is 1 (that's an add or delete),
        2 if both are > 1 (delete then add)
    style diff: 1

chordsym symbol count:
    1 symbol for chord name
    1 symbol if styled in any way

chordsym diff symbolic edit count:
    chord name diff: 2 (delete then add)
    style diff: 1

ending symbol count:
    len(string) symbols for ending name
    1 symbol for measure count

ending diff symbolic edit count:
    name diff: Levenshtein distance between strings
    measure count diff: 2 (delete then add)

system break/page break symbol count:
    1 symbol

system break/page break diff symbolic edit count:
    they are never different, just deleted/inserted, with symbolic edit count == 1

staffinfo symbol count:
    1 symbol if lines per staff specified
    1 symbol if staff size % specified

staffinfo diff symbolic edit count:
    lines/staff diff: 2 (delete then add)
    staff size diff: 2 (delete then add)

-------------------------------------------------------------------

AnnLyric symbol count (a single word or syllable):
    len(text) symbols for word/syllable (including any implied hyphens, etc)
    1 symbol for offset (horizontal position in measure)
    1 symbol if there is a verse number
    1 symbol if there is a verse identifier (that is not the verse number)
    1 symbol if styled (text style, color, etc)

AnnLyric diff symbolic edit count:
    text diff: Levenshtein distance between strings
    verse number diff: 2 (delete then add)
    verse identifier diff: 1 if delete or add, 2 if verse identifier changed
        (delete then add)
    offset diff: 1 (horizontal shift)
    style diff: 1

-------------------------------------------------------------------

AnnVoice symbol count (if Voicing is not explicitly specified on the command line,
    there are no AnnVoices):
    Sum of the symbols in all the AnnNotes in the AnnVoice

AnnVoice diff symbolic edit count:
    Sum of the symbolic edits due to the AnnNotes in the AnnVoice
    If a voice is deleted or added, the symbolic edit count is the AnnVoice
        symbol count.

-------------------------------------------------------------------

AnnMeasure symbol count:
    if Voicing specified explicitly:
        Sum of symbol count for all AnnVoices, AnnExtras, and AnnLyrics in the AnnMeasure
    Normally:
        Sum of symbol count for all AnnNotes, AnnExtras, and AnnLyrics in the AnnMeasure

AnnMeasure diff symbolic edit count:
    Sum of the symbolic edits for all objects in the AnnMeasure.
    If Voicing specified explicitly:
        If an entire AnnVoice is deleted or added, the symbolic edit count is
            incremented by that AnnVoice's symbol count.

-------------------------------------------------------------------

AnnPart symbol count:
    Sum of symbol count for all AnnMeasures in the AnnPart

AnnPart diff symbolic edit count:
    Sum of the symbolic edits for all AnnMeasures in the AnnPart.
    If an entire AnnMeasure is deleted or added, the symbolic edit count is
        incremented by that AnnMeasure's symbol count.

-------------------------------------------------------------------

AnnStaffGroup symbol count:
    len(name) symbols for staff group name (e.g. instrument name)
    len(abbrev) symbols for staff group abbreviation
    1 symbol for staff group symbol shape (curly brace, bracket, line, etc)
    1 symbol for staff group barline type (bar together, don't bar together,
        Mensurstrich)
    1 symbol for the lowest staff index the staff group encloses (i.e.
        vertical position)
    1 symbol for the highest staff index the staff group encloses (i.e.
        vertical height)

AnnStaffGroup diff symbolic edit count:
    name diff: Levenshtein distance between names
    abbreviation diff: Levenshtein distance between abbreviations
    symbol diff: 1 if adding or deleting symbol, 2 if changing symbol (delete
        then add)
    barline type diff: 1
    staff list diff: 1 for diff in lowest staff index (vertical shift of group
        top), 1 for diff in highest staff index (vertical shift of group bottom)

-------------------------------------------------------------------

AnnMetadataItem symbol count (e.g. composer name, date composed, etc):
    len(key) + len(value) symbols for the item

AnnMetadataItem diff symbolic edit count:
    key difference: We only compare items that have the same key. Any items
        with no same-keyed item to compare against are deleted or added (with
        symbolic edit cost == item symbol count)
    value difference: Levenshtein distance between value strings

-------------------------------------------------------------------

AnnScore symbol count:
    Sum of symbol counts for all AnnParts, AnnStaffGroups, and AnnMetadataItems in the AnnScore.

AnnScore diff symbolic edit count:
    Sum of symbolic edits in all AnnParts, AnnStaffGroups, and AnnMetadataItems
        in the AnnScore.
    If an entire AnnPart is added or deleted, the symbolic edit count is incremented
        by that AnnPart's symbol count.