# ------------------------------------------------------------------------------
# Name: reduction.py
# Purpose: Tools for creating a score reduction.
#
# Authors: Christopher Ariza
# Michael Scott Asato Cuthbert
#
# Copyright: Copyright © 2011-2013 Michael Scott Asato Cuthbert
# License: BSD, see license.txt
# ------------------------------------------------------------------------------
'''
Tools for generation reduction displays, showing a score and or a chord reduction,
and one or more reductive representation lines.
Used by graph.PlotHorizontalBarWeighted()
'''
from __future__ import annotations
import copy
import re
import typing as t
import unittest
from music21 import exceptions21
from music21 import chord
from music21 import common
from music21.common.types import DocOrder
from music21 import environment
from music21 import expressions
from music21 import instrument
from music21 import note
from music21 import pitch
from music21 import prebase
from music21 import stream
environLocal = environment.Environment('analysis.reduction')
# ------------------------------------------------------------------------------
class ReductiveEventException(exceptions21.Music21Exception):
pass
# as lyric, or as parameter
#
# ::/p:g#/o:5/nh:f/ns:n/l:1/g:ursatz/v:1
[docs]
class ReductiveNote(prebase.ProtoM21Object):
'''
The extraction of an event from a score and specification of where
and how it should be presented in a reductive score.
A specification string, as well as Note, must be provided for parsing.
A specification must be created when access the Measure that the source note
is found in. Storing the measure and index position provides significant
performance optimization, as we do no have to search
every note when generated the reduction.
The `measureIndex` is the index of measure where this is found, not
the measure number. The `measureOffset` is the position in the measure
specified by the index.
'''
_delimitValue = ':' # store the delimiter string, must start with 2
_delimitArg = '/'
# map the abbreviation to the data key
_parameterKeys = {
'p': 'pitch',
'o': 'octave',
'nf': 'noteheadFill',
'sd': 'stemDirection',
'g': 'group',
'v': 'voice',
'ta': 'textAbove', # text annotation
'tb': 'textBelow', # text annotation
}
_defaultParameters = {
'pitch': None, # use notes, or if a chord take highest
'octave': None, # use notes
'noteheadFill': None, # use notes
'stemDirection': 'noStem',
'group': None,
'voice': None,
}
def __init__(self, specification, inputNote, measureIndex, measureOffset):
self._specification = specification
self._note = None # store a reference to the note this is attached to
self._parameters = {}
# do parsing if possible
self._isParsed = False
self._parseSpecification(self._specification)
self._note = inputNote # keep a reference
self.measureIndex = measureIndex
self.measureOffset = measureOffset
def _reprInternal(self):
msg = []
for key in self._parameterKeys:
attr = self._parameterKeys[key]
if attr in self._parameters: # only show those defined
if self._parameters[attr]:
msg.append(key)
msg.append(':')
msg.append(self._parameters[attr])
if self._note is not None:
msg.append(' of ')
msg.append(repr(self._note))
return ''.join(msg)
[docs]
def __getitem__(self, key):
return self._parameters[key]
def _parseSpecification(self, spec: str):
# start with the defaults
self._parameters = copy.deepcopy(self._defaultParameters)
spec = spec.strip()
# spec = spec.replace(' ', '')
if not spec.startswith(self._delimitValue + self._delimitValue):
return # nothing to parse
args = spec.split(self._delimitArg)
for a in args[1:]: # skip the first arg, as it is just delimiter
# if no delimit arg, it cannot be parsed
if self._delimitValue not in a:
continue
candidateKey, value = a.split(self._delimitValue)
candidateKey = candidateKey.strip()
value = value.strip()
if candidateKey.lower() in self._parameterKeys:
attr = self._parameterKeys[candidateKey]
self._parameters[attr] = value
self._isParsed = True
[docs]
def isParsed(self) -> bool:
return self._isParsed
[docs]
def getNoteAndTextExpression(self):
'''
Produce a new note, a deep copy of the supplied note
and with the specified modifications.
'''
n = None
if self._note.isChord:
# need to permit specification by pitch
if 'pitch' in self._parameters:
p = pitch.Pitch(self._parameters['pitch'])
for sub in self._note: # iterate over components
if p.name.lower() == sub.pitch.name.lower():
# copy the component
n = copy.deepcopy(sub)
else: # get first, or get entire chord?
# n = copy.deepcopy(self._note.pitches[0])
n = copy.deepcopy(self._note.pitches[0])
else:
n = copy.deepcopy(self._note)
# always clear certain parameters
if n is None:
pitchParameter = self._parameters['pitch']
raise ReductiveEventException(
f'Could not find pitch, {pitchParameter!r} in self._note: {self._note!r}')
n.lyrics = []
n.tie = None
n.expressions = []
n.articulations = []
n.duration.dots = 0 # set to zero
if n.pitch.accidental is not None:
n.pitch.accidental.displayStatus = True
te = None
if 'octave' in self._parameters:
if self._parameters['octave']:
n.pitch.octave = self._parameters['octave']
if 'stemDirection' in self._parameters:
n.stemDirection = self._parameters['stemDirection']
if 'noteheadFill' in self._parameters:
nhf = self._parameters['noteheadFill']
if nhf:
if nhf == 'yes':
nhf = True
elif nhf == 'no':
nhf = False
n.noteheadFill = nhf
# environLocal.printDebug(['set notehead fill:', n.noteheadFill])
if 'textBelow' in self._parameters:
n.addLyric(self._parameters['textBelow'])
if 'textAbove' in self._parameters:
te = expressions.TextExpression(self._parameters['textAbove'])
return n, te
# ------------------------------------------------------------------------------
class ScoreReductionException(exceptions21.Music21Exception):
pass
[docs]
class ScoreReduction:
'''
An object to reduce a score.
'''
def __init__(self, **keywords):
# store a list of one or more reductions
self._reductiveNotes = {}
self._reductiveVoices = []
self._reductiveGroups = []
# store the source score
self._score = None
self._chordReduction = None # store a chordal reduction of available
def _setScore(self, value):
if not isinstance(value, stream.Stream):
raise ScoreReductionException('cannot set a non Stream')
if value.hasPartLikeStreams:
# make a local copy
self._score = copy.deepcopy(value)
else: # assume a single stream, place in a Score
s = stream.Score()
s.insert(0, copy.deepcopy(value))
self._score = s
self._score.setDerivationMethod('ScoreReduction', recurse=True)
def _getScore(self):
return self._score
score = property(_getScore, _setScore, doc='''
Get or set the Score. Setting the score set a deepcopy of the score; the score
set here will not be altered.
>>> s = corpus.parse('bwv66.6')
>>> sr = analysis.reduction.ScoreReduction()
>>> sr.score = s
''')
def _setChordReduction(self, value):
if not isinstance(value, stream.Stream):
raise ScoreReductionException('cannot set a non Stream')
if value.hasPartLikeStreams():
# make a local copy
self._chordReduction = copy.deepcopy(value)
else: # assume a single stream, place in a Score
s = stream.Score()
s.insert(0, copy.deepcopy(value))
self._chordReduction = s
def _getChordReduction(self):
return self._chordReduction
chordReduction = property(_getChordReduction, _setChordReduction, doc='''
Get or set a Chord reduction as a Stream or Score. Setting the this values
set a deepcopy of the reduction; the reduction set here will not be altered.
''')
def _extractReductionEvents(self, score, removeAfterParsing=True):
'''
Remove and store all reductive events
Store in a dictionary where obj id is obj key
'''
if score is None:
return
# iterate overall notes, check all lyrics
for p in score.parts:
for i, m in enumerate(p.getElementsByClass(stream.Measure)):
for n in m.recurse().notes:
infoDict = {'part': p,
'measure': m,
'measureIndex': i}
self._extractNoteReductiveEvent(n, infoDict, removeAfterParsing)
def _extractNoteReductiveEvent(self, n, infoDict=None, removeAfterParsing=True):
if infoDict is None:
infoDict = {'part': None,
'measure': None,
'measureIndex': 0
}
m = infoDict['measure']
if not n.lyrics:
return
removalIndices = []
if n in m:
offset = n.getOffsetBySite(m)
else: # it is in a Voice
offset = 0.0
for v in m.voices:
if n in v:
offset = n.getOffsetBySite(v)
# a list of Lyric objects
for k, lyr in enumerate(n.lyrics):
# store measure index
rn = ReductiveNote(lyr.text, n, infoDict['measureIndex'], offset)
if rn.isParsed():
# environLocal.printDebug(['parsing reductive note', rn])
# use id, lyric text as hash
key = str(id(n)) + lyr.text
self._reductiveNotes[key] = rn
removalIndices.append(k)
if removeAfterParsing:
for q in removalIndices:
# replace position in list with empty lyric
n.lyrics[q] = note.Lyric('')
def _parseReductiveNotes(self):
self._reductiveNotes = {}
self._extractReductionEvents(self._chordReduction)
self._extractReductionEvents(self._score)
for unused_key, rn in self._reductiveNotes.items():
if rn['group'] not in self._reductiveGroups:
self._reductiveGroups.append(rn['group'])
if rn['voice'] not in self._reductiveVoices:
self._reductiveVoices.append(rn['voice'])
# if we have None and a group, then we should just use that one
# group; same with voices
if (len(self._reductiveGroups) == 2
and None in self._reductiveGroups):
self._reductiveGroups.remove(None)
# for now, sort all
# environLocal.printDebug(['self._reductiveGroups', self._reductiveGroups])
if (len(self._reductiveVoices) == 2
and None in self._reductiveVoices):
self._reductiveVoices.remove(None)
def _createReduction(self):
self._parseReductiveNotes()
s = stream.Score()
# need to scan all tags
oneGroup = False
if len(self._reductiveGroups) == 1:
# if 1, can be None or a group name:
oneGroup = True
oneVoice = False
if len(self._reductiveVoices) == 1:
# if 1, can be None or a group name:
oneVoice = True
if self._score:
mTemplate = self._score.parts.first().template(retainVoices=False)
else:
mTemplate = self._chordReduction.parts.first().template(retainVoices=False)
# for each defined reductive group
for gName in self._reductiveGroups:
# create reductive parts
# need to break by necessary parts, voices; for now, assume one
g = copy.deepcopy(mTemplate)
g.id = gName
inst = instrument.Instrument()
inst.partName = gName
g.insert(0, inst)
gMeasures = g.getElementsByClass(stream.Measure)
# for m in gMeasures._elements:
# print(gName, m)
# m.clef = clef.TrebleClef()
# TODO: insert into note or chord
for unused_key, rn in self._reductiveNotes.items():
if oneGroup or rn['group'] == gName:
# environLocal.printDebug([
# '_createReduction(): found reductive note, rn', rn, 'group', gName])
gMeasure = gMeasures[rn.measureIndex]
if not gMeasure.voices: # common setup routines
# if no voices, start by removing rests
gMeasure.removeByClass('Rest')
for vId in self._reductiveVoices:
v = stream.Voice()
v.id = vId
gMeasure.insert(0, v)
if oneVoice:
n, te = rn.getNoteAndTextExpression()
gMeasure.voices[0].insertIntoNoteOrChord(
rn.measureOffset, n)
# place the text expression in the Measure, not Voice
if te:
gMeasure.insert(rn.measureOffset, te)
else:
v = gMeasure.getElementById(rn['voice'])
if v is None: # just take the first
v = gMeasure.voices[0]
n, te = rn.getNoteAndTextExpression()
v.insertIntoNoteOrChord(rn.measureOffset, n)
if te:
gMeasure.insert(rn.measureOffset, te)
# after gathering all parts, fill with rests
for i, m in enumerate(g.getElementsByClass(stream.Measure)):
# only make rests if there are notes in the measure
for v in m.voices:
if v.recurse().notes:
v.makeRests(fillGaps=True, inPlace=True)
m.flattenUnnecessaryVoices(inPlace=True)
# hide all rests in all containers
for r in m[note.Rest]:
r.style.hideObjectOnPrint = True
# m.show('t')
# add to score
s.insert(0, g)
# g.show('t')
if self._chordReduction:
for p in self._chordReduction.parts:
s.insert(0, p)
srcParts = [] # for bracket
if self._score:
for p in self._score.parts:
s.insert(0, p)
srcParts.append(p) # store to brace
return s
[docs]
def reduce(self):
'''
Given a score, populate this Score reduction
'''
# if not set here or before
if self.score is None and self.chordReduction is None:
raise ScoreReductionException('no score defined to reduce')
return self._createReduction()
# ------------------------------------------------------------------------------
class PartReductionException(exceptions21.Music21Exception):
pass
# ------------------------------------------------------------------------------
[docs]
class PartReduction:
'''
A part reduction reduces a Score into one or more parts.
Parts are combined based on a part group dictionary.
Each resulting part is then segmented by an object.
This object is assigned as floating-point value.
This reduction is designed to work with the GraphHorizontalBarWeighted and related Plot
subclasses.
If the `fillByMeasure` parameter is True, and if measures are available,
each part will segment by Measure divisions, and look for the target activity only
once per Measure.
If more than one target is found in the Measure, values will be averaged.
If `fillByMeasure` is False, the part will be segmented by each Note.
The `segmentByTarget` parameter is True, segments, which may be Notes or Measures,
will be divided if necessary to show changes that occur over the duration of the
segment by a target object.
If the `normalizeByPart` parameter is True, each part will be normalized within
the range only of that part. If False, all parts will be normalized by the max
of all parts. The default is True.
If the `normalize` parameter is False, no normalization will take place. The default is True.
'''
def __init__(self,
srcScore=None,
*,
partGroups: list[dict[str, t.Any]]|None = None,
fillByMeasure: bool = True,
segmentByTarget: bool = True,
normalize: bool = True,
normalizeByPart: bool = False,
**keywords):
if srcScore is None:
return
if not isinstance(srcScore, stream.Score):
raise PartReductionException('provided Stream must be Score')
self._score = srcScore
# an ordered list of dictionaries for
# part id, part color, and a list of Part objs
# TODO: typed dict
self._partBundles: list[dict[str, t.Any]] = []
# a dictionary of part id to a list of events
self._eventSpans: dict[str|int, list[t.Any]] = {}
# define how parts are grouped
# a list of dictionaries, with keys for name, color, and a match list
self._partGroups = partGroups
self._fillByMeasure = fillByMeasure
# We re-partition if the spans change
self._segmentByTarget = segmentByTarget
self._normalizeByPart = normalizeByPart # norm by all parts is default
self._normalizeToggle = normalize
# check that there are measures
for p in self._score.parts:
if not p.hasMeasures():
self._fillByMeasure = False
# environLocal.printDebug(['overriding fillByMeasure as no measures are defined'])
break
def _createPartBundles(self):
'''
Fill the _partBundles list with dictionaries,
each dictionary defining a name (part id or supplied), a color, and list
of Parts that match.
'''
self._partBundles = []
if self._partGroups:
for d in self._partGroups: # a list of dictionaries
name, pColor, matches = d['name'], d['color'], d['match']
sub = []
for p in self._score.parts:
# environLocal.printDebug(['_createPartBundles: part.id', p.id])
# if matches is None, use group name
if matches is None:
matches = [name]
pId = str(p.id).lower()
for m in matches: # strings or instruments
if (isinstance(m, str)
and pId.find(m.lower()) >= 0):
sub.append(p)
break
elif re.match(m.lower(), pId):
sub.append(p)
# TODO: match if m is Instrument class
if not sub:
continue
data = {
'pGroupId': name,
'color': pColor,
'parts': sub,
}
self._partBundles.append(data)
else: # manually creates
for p in self._score.parts:
# store one or more Parts associated with an id
data = {'pGroupId': p.id, 'color': '#666666', 'parts': [p]}
self._partBundles.append(data)
# create flat representation of all parts in a bundle
for partBundle in self._partBundles:
if len(partBundle['parts']) == 1:
partBundle['parts.flat'] = partBundle['parts'][0].flatten()
else:
# align all parts and flatten
# this takes a flat presentation of all parts
s = stream.Stream()
for p in partBundle['parts']:
s.insert(0, p)
partBundle['parts.flat'] = s.flatten()
def _createEventSpans(self):
# for each part group id key, store a list of events
self._eventSpans = {}
for partBundle in self._partBundles:
pGroupId = partBundle['pGroupId']
pColor = partBundle['color']
parts = partBundle['parts']
# print(pGroupId)
dataEvents = []
# combine multiple streams into a single
eStart = None
eEnd = None
eLast = None
# segmenting by measure if that measure contains notes.
# Note that measures are not elided of activity is contiguous
if self._fillByMeasure:
partMeasures = []
for p in parts:
partMeasures.append(p.getElementsByClass(stream.Measure).stream())
# environLocal.printDebug(['partMeasures', partMeasures])
# assuming that all parts have same the number of measures
# iterate over each measure
# iLast = len(partMeasures[0]) - 1
for i in range(len(partMeasures[0])):
active = False
# check for activity in any part in the part group
for p in partMeasures: # iter of parts containing measures
# print(p, i, p[i], len(p[i].flatten().notes))
if p[i].iter().notes:
active = True
break
# environLocal.printDebug([i, 'active', active])
if not active:
continue
# get offset, or start, of this measure
e = partMeasures[0][i]
eStart = e.getOffsetBySite(partMeasures[0])
# use duration, not barDuration.quarterLength
# as want filled duration?
eEnd = (eStart + e.barDuration.quarterLength)
ds = {'eStart': eStart,
'span': eEnd - eStart,
'weight': None,
'color': pColor,
}
dataEvents.append(ds)
# if eStart is None and active:
# eStart = partMeasures[0][i].getOffsetBySite(
# partMeasures[0])
# elif (eStart is not None and not active) or i >= iLast:
# if eStart is None: # nothing to do; just the last
# continue
# # if this is the last measure, and it is active
# if (i >= iLast and active):
# eLast = partMeasures[0][i]
# # use duration, not barDuration.quarterLength
# # as want filled duration?
# eEnd = (eLast.getOffsetBySite(partMeasures[0])
# + eLast.barDuration.quarterLength)
# ds = {'eStart': eStart, 'span': eEnd - eStart,
# 'weight': None, 'color': pColor}
# dataEvents.append(ds)
# eStart = None
# eLast = partMeasures[0][i]
# fill by alternative approach, based on activity of notes
# creates region for each contiguous span of notes
# this is useful as it will handle overlaps and similar arrangements
# TODO: this needs further testing
else:
# this takes a flat presentation of all parts, and then
# finds any gaps in consecutive notes
eSrc = partBundle['parts.flat']
# a li=st, not a stream
# a None in the resulting list designates a rest
noteSrc = eSrc.findConsecutiveNotes()
for i, e in enumerate(noteSrc):
# environLocal.printDebug(['i, e', i, e])
# if this event is a rest, e is None
if e is None:
if eStart is None: # the first event is a rest
continue
else:
eEnd = eLast.getOffsetBySite(eSrc) + eLast.quarterLength
# create a temporary weight
ds = {'eStart': eStart,
'span': eEnd - eStart,
'weight': None,
'color': pColor,
}
dataEvents.append(ds)
eStart = None
elif i >= len(noteSrc) - 1: # this is the last
if eStart is None: # the last event was a rest
# this the start is the start of this event
eStart = e.getOffsetBySite(eSrc)
eEnd = e.getOffsetBySite(eSrc) + e.quarterLength
# create a temporary weight
ds = {'eStart': eStart,
'span': eEnd - eStart,
'weight': None,
'color': pColor,
}
dataEvents.append(ds)
eStart = None
else:
if eStart is None:
eStart = e.getOffsetBySite(eSrc)
eLast = e
# environLocal.printDebug(['dataEvents', dataEvents])
self._eventSpans[pGroupId] = dataEvents
def _getValueForSpan(
self,
target='Dynamic',
splitSpans=True,
targetToWeight=None
):
'''
For each span, determine the measured parameter value. This is translated
as the height of the bar graph.
If `splitSpans` is True, a span will be split of the target changes over the span.
Otherwise, Spans will be averaged. This is the `segmentByTarget` parameter.
The `targetToWeight` parameter is a function that takes a list or Stream of objects
(of the class specified by `target`) and returns a single floating-point value.
'''
# this temporary function only works with dynamics
def _dynamicToWeight(targets):
# permit a stream
if hasattr(targets, 'isStream') and targets.isStream:
pass
elif not common.isIterable(targets):
targets = [targets]
summation = 0
for e in targets: # a Stream
summation += e.volumeScalar # for dynamics
return summation / len(target)
# supply function to convert one or more targets to number
if targetToWeight is None:
targetToWeight = _dynamicToWeight
if not splitSpans: # this is segmentByTarget
for partBundle in self._partBundles:
flatRef = partBundle['parts.flat']
for ds in self._eventSpans[partBundle['pGroupId']]:
# for each event span, find the targeted object
offsetStart = ds['eStart']
offsetEnd = offsetStart + ds['span']
match = flatRef.getElementsByOffset(
offsetStart,
offsetEnd,
includeEndBoundary=False,
mustFinishInSpan=False,
mustBeginInSpan=True
).getElementsByClass(target).stream()
if not match:
w = None
else:
w = targetToWeight(match)
# environLocal.printDebug(['segment weight', w])
ds['weight'] = w
else:
for partBundle in self._partBundles:
finalBundle = []
flatRef = partBundle['parts.flat']
# get each span
for ds in self._eventSpans[partBundle['pGroupId']]:
offsetStart = ds['eStart']
offsetEnd = offsetStart + ds['span']
# get all targets within the contiguous region
# e.g., Dynamics objects
match = flatRef.getElementsByOffset(offsetStart, offsetEnd,
includeEndBoundary=True, mustFinishInSpan=False,
mustBeginInSpan=True).getElementsByClass(target).stream()
# environLocal.printDebug(['matched elements', target, match])
# extend duration of all found dynamics
match.extendDuration(target, inPlace=True)
# match.show('t')
dsFirst = copy.deepcopy(ds)
if not match:
# weight is not known
finalBundle.append(dsFirst)
continue
# create new spans for each target in this segment
for i, tar in enumerate(match):
targetStart = tar.getOffsetBySite(flatRef)
# can use extended duration
targetSpan = tar.duration.quarterLength
# if dur of target is greater tn this span
# end at this span
if targetStart + targetSpan > offsetEnd:
targetSpan = offsetEnd - targetStart
# if we have the last matched target, it will
# have zero duration, as there is no following
# thus, span needs to be distance to end of regions
if targetSpan <= 0.001:
targetSpan = offsetEnd - targetStart
# environLocal.printDebug([t, 'targetSpan', targetSpan,
# 'offsetEnd', offsetEnd, "ds['span']", ds['span']])
if i == 0 and ds['eStart'] == targetStart:
# the target start at the same position
# as the start of this existing span
# dsFirst['eStart'] = targetStart
dsFirst['span'] = targetSpan
dsFirst['weight'] = targetToWeight(tar)
finalBundle.append(dsFirst)
elif t == 0 and ds['eStart'] != targetStart:
# add two, one for the empty region, one for target
# adjust span of first; weight is not known
# (hangs over from last)
dsFirst['span'] = targetStart - offsetStart
finalBundle.append(dsFirst)
dsNext = copy.deepcopy(ds)
dsNext['eStart'] = targetStart
dsNext['span'] = targetSpan
dsNext['weight'] = targetToWeight(tar)
finalBundle.append(dsNext)
else: # for all other cases, create segment for each
dsNext = copy.deepcopy(ds)
dsNext['eStart'] = targetStart
dsNext['span'] = targetSpan
dsNext['weight'] = targetToWeight(tar)
finalBundle.append(dsNext)
# after iterating all ds spans, reassign
self._eventSpans[partBundle['pGroupId']] = finalBundle
def _extendSpans(self):
'''
Extend the value of a target parameter to the next boundary.
An undefined boundary will wave as its weight None.
'''
# environLocal.printDebug(['_extendSpans: pre'])
# for partBundle in self._partBundles:
# for i, ds in enumerate(self._eventSpans[partBundle['pGroupId']]):
# print(ds)
minValue = 0.01 # for error conditions
for partBundle in self._partBundles:
lastWeight = None
for i, ds in enumerate(self._eventSpans[partBundle['pGroupId']]):
if i == 0: # cannot extend first
if ds['weight'] is None: # this is an error in the rep
ds['weight'] = minValue
# environLocal.printDebug([
# 'cannot extend a weight: no previous weight defined'])
else:
lastWeight = ds['weight']
else: # not first
if ds['weight']:
lastWeight = ds['weight']
elif lastWeight: # its None, use last
ds['weight'] = lastWeight
# do not have a list; mist set to min
elif ds['weight'] is None and lastWeight is None:
ds['weight'] = minValue
# environLocal.printDebug([
# 'cannot extend a weight: no previous weight defined'])
# environLocal.printDebug(['_extendSpans: post'])
# for partBundle in self._partBundles:
# for i, ds in enumerate(self._eventSpans[partBundle['pGroupId']]):
# print(ds)
def _normalize(self, byPart=False):
'''
Normalize, either within each Part, or for all parts
'''
partMaxRef = {}
for partBundle in self._partBundles:
partMax = 0
for ds in self._eventSpans[partBundle['pGroupId']]:
if ds['weight'] > partMax:
partMax = ds['weight']
partMaxRef[partBundle['pGroupId']] = partMax
try:
maxOfMax = max(partMaxRef.values())
except ValueError: # empty part?
maxOfMax = 0
for partBundle in self._partBundles:
for ds in self._eventSpans[partBundle['pGroupId']]:
# weight is now fraction of the max for that part
if byPart:
bestMax = partMaxRef[partBundle['pGroupId']]
else:
bestMax = maxOfMax
if bestMax != 0:
ds['weight'] = (ds['weight'] / bestMax)
else:
ds['weight'] = 1 # error?
[docs]
def process(self):
'''
Core processing routines.
'''
self._createPartBundles()
self._createEventSpans()
self._getValueForSpan(splitSpans=self._segmentByTarget)
self._extendSpans()
if self._normalizeToggle:
self._normalize(byPart=self._normalizeByPart)
[docs]
def getGraphHorizontalBarWeightedData(self):
'''
Get all data organized into bar span specifications.
'''
# data = [
# ('Violins', [(3, 5, 1, '#fff000'), (1, 12, 0.2, '#3ff203', 0.1, 1)] ),
# ('Celli', [(2, 7, 0.2, '#0ff302'), (10, 3, 0.6, '#ff0000', 1)] ),
# ]
data = []
# iterate over part bundles to get order
for partBundle in self._partBundles:
# print(partBundle)
dataList = []
groupSpans = partBundle['pGroupId']
for ds in self._eventSpans[groupSpans]:
# data format here is set by the graphing routine
dataList.append([ds['eStart'], ds['span'], ds['weight'], ds['color']])
data.append((partBundle['pGroupId'], dataList))
return data
# ------------------------------------------------------------------------------
class Test(unittest.TestCase):
def testCopyAndDeepcopy(self):
from music21.test.commonTest import testCopyAll
testCopyAll(self, globals())
def testExtractionA(self):
from music21 import analysis
from music21 import corpus
s = corpus.parse('bwv66.6')
# s.show()
s.parts[0].flatten().notes[3].addLyric('test')
s.parts[0].flatten().notes[4].addLyric('::/o:6/tb:here')
s.parts[3].flatten().notes[2].addLyric('::/o:5/tb:fromBass')
s.parts[1].flatten().notes[7].addLyric('::/o:4/nf:no/g:Ursatz/ta:3 3 200')
sr = analysis.reduction.ScoreReduction()
sr.score = s
post = sr.reduce()
# post.show()
# post.parts[0].show('t')
self.assertEqual(len(post.parts[0].flatten().notes), 3)
# post.parts[0].show('t')
three_measures = post.parts.first()[stream.Measure][:3]
new_stream = stream.Stream()
for m in three_measures:
new_stream.append(m)
flat_stream = new_stream.flatten()
match = [(repr(e), e.offset, e.duration.quarterLength) for e in flat_stream.notesAndRests]
self.maxDiff = None
self.assertEqual(match,
[('<music21.note.Rest quarter>', 0.0, 1.0),
('<music21.note.Note F#>', 1.0, 1.0),
('<music21.note.Rest quarter>', 2.0, 1.0),
('<music21.note.Note C#>', 3.0, 1.0),
('<music21.note.Rest quarter>', 4.0, 1.0),
('<music21.note.Note G#>', 5.0, 1.0)])
# test that lyric is found
self.assertEqual(post.parts[0].flatten().notes[0].lyric, 'fromBass')
def testExtractionB(self):
from music21 import analysis
from music21 import corpus
s = corpus.parse('bwv66.6')
s.parts[0].flatten().notes[4].addLyric('::/o:6/v:1/tb:s/g:Ursatz')
s.parts[3].flatten().notes[2].addLyric('::/o:5/v:2/tb:b')
s.parts[2].flatten().notes[3].addLyric('::/o:4/v:2/tb:t')
s.parts[1].flatten().notes[2].addLyric('::/o:4/v:2/tb:a')
sr = analysis.reduction.ScoreReduction()
extract = s.measures(0, 10)
# extract.show()
sr.score = extract
# sr.score = s
post = sr.reduce()
# post.show()
self.assertEqual(len(post.parts), 5)
match = post.parts[0].flatten().notes
self.assertEqual(len(match), 3)
# post.show()
# def testExtractionC(self):
# from music21 import analysis
# from music21 import corpus
# # http://solomonsmusic.net/schenker.htm
# # shows extracting an Ursatz line
#
# # BACH pre;ide !, WTC
#
# src = corpus.parse('bwv846')
# import warnings
# with warnings.catch_warnings(): # catch deprecation warning
# warnings.simplefilter('ignore', category=exceptions21.Music21DeprecationWarning)
# chords = src.flatten().makeChords(minimumWindowSize=4, # make chords is gone
# makeRests=False)
# for c in chords.flatten().notes:
# c.quarterLength = 4
# for m in chords.getElementsByClass(stream.Measure):
# m.clef = clef.bestClef(m, recurse=True)
#
# chords.measure(1).notes[0].addLyric('::/p:e/o:5/nf:no/ta:3/g:Ursatz')
# chords.measure(1).notes[0].addLyric('::/p:c/o:4/nf:no/tb:I')
#
# chords.measure(24).notes[0].addLyric('::/p:d/o:5/nf:no/ta:2')
# chords.measure(24).notes[0].addLyric('::/p:g/o:3/nf:no/tb:V')
#
# chords.measure(30).notes[0].addLyric('::/p:f/o:4/tb:7')
#
# chords.measure(34).notes[0].addLyric('::/p:c/o:5/nf:no/v:1/ta:1')
# chords.measure(34).notes[0].addLyric('::/p:g/o:4/nf:no/v:2')
# chords.measure(34).notes[0].addLyric('::/p:c/o:4/nf:no/v:1/tb:I')
#
# sr = analysis.reduction.ScoreReduction()
# sr.chordReduction = chords
# # sr.score = src
# unused_post = sr.reduce()
# # unused_post.show()
def testExtractionD(self):
# this shows a score, extracting a single pitch
from music21 import analysis
from music21 import corpus
src = corpus.parse('schoenberg/opus19', 6)
for n in src.flatten().notes:
if isinstance(n, note.Note):
if n.pitch.name == 'F#':
n.addLyric('::/p:f#/o:4')
# if n.pitch.name == 'C':
# n.addLyric('::/p:c/o:4/g:C')
elif isinstance(n, chord.Chord):
if 'F#' in [p.name for p in n.pitches]:
n.addLyric('::/p:f#/o:4')
# if 'C' in [p.name for p in n.pitches]:
# n.addLyric('::/p:c/o:4/g:C')
sr = analysis.reduction.ScoreReduction()
sr.score = src
unused_post = sr.reduce()
# post.show()
def testExtractionD2(self):
# this shows a score, extracting a single pitch
from music21 import analysis
from music21 import corpus
src = corpus.parse('schoenberg/opus19', 6)
for n in src.flatten().notes:
if isinstance(n, note.Note):
if n.pitch.name == 'F#':
n.addLyric('::/p:f#/o:4/g:F#')
if n.pitch.name == 'C':
n.addLyric('::/p:c/o:4/g:C')
elif isinstance(n, chord.Chord):
if 'F#' in [p.name for p in n.pitches]:
n.addLyric('::/p:f#/o:4/g:F#')
if 'C' in [p.name for p in n.pitches]:
n.addLyric('::/p:c/o:4/g:C')
sr = analysis.reduction.ScoreReduction()
sr.score = src
unused_post = sr.reduce()
# post.show()
def testExtractionE(self):
from music21 import analysis
from music21 import corpus
src = corpus.parse('corelli/opus3no1/1grave')
# chords = src.chordify()
sr = analysis.reduction.ScoreReduction()
# sr.chordReduction = chords
sr.score = src
unused_post = sr.reduce()
# post.show()
def testPartReductionA(self):
from music21 import analysis
from music21 import corpus
s = corpus.parse('bwv66.6')
partGroups = [
{
'name': 'High Voices',
'color': '#ff0088',
'match': ['soprano', 'alto']
},
{
'name': 'Low Voices',
'color': '#8800ff',
'match': ['tenor', 'bass']
},
]
pr = analysis.reduction.PartReduction(s, partGroups=partGroups)
pr.process()
for sub in pr._partGroups:
self.assertEqual(len(sub['match']), 2)
def _matchWeightedData(self, match, target):
'''
Utility function to compare known data but not compare floating point weights.
'''
for partId, b in enumerate(target):
a = match[partId]
self.assertEqual(a[0], b[0])
for i, dataMatch in enumerate(a[1]): # second item has data
dataTarget = b[1][i]
# start
self.assertAlmostEqual(dataMatch[0], dataTarget[0])
# span
self.assertAlmostEqual(dataMatch[1], dataTarget[1])
# weight
self.assertAlmostEqual(
dataMatch[2],
dataTarget[2],
msg=(f'for partId {partId}, entry {i}: '
f'should be {dataMatch[2]} <-> was {dataTarget[2]}')
)
def testPartReductionB(self, show=False):
'''
Artificially create test cases.
'''
from music21 import analysis
from music21 import dynamics
from music21 import graph
durDynPairsA = [(1, 'mf'), (3, 'f'), (2, 'p'), (4, 'ff'), (2, 'mf')]
durDynPairsB = [(1, 'mf'), (3, 'f'), (2, 'p'), (4, 'ff'), (2, 'mf')]
s = stream.Score()
pCount = 0
for pairs in [durDynPairsA, durDynPairsB]:
p = stream.Part()
p.id = pCount
pos = 0
for ql, dyn in pairs:
p.insert(pos, note.Note(quarterLength=ql))
p.insert(pos, dynamics.Dynamic(dyn))
pos += ql
# p.makeMeasures(inPlace=True)
s.insert(0, p)
pCount += 1
if show is True:
s.show()
pr = analysis.reduction.PartReduction(s, normalize=False)
pr.process()
match = pr.getGraphHorizontalBarWeightedData()
target = [(0, [[0.0, 1.0, 0.07857142857142858, '#666666'],
[1.0, 3.0, 0.09999999999999999, '#666666'],
[4.0, 2.0, 0.05, '#666666'],
[6.0, 4.0, 0.12142857142857143, '#666666'],
[10.0, 2.0, 0.07857142857142858, '#666666']]),
(1, [[0.0, 1.0, 0.07857142857142858, '#666666'],
[1.0, 3.0, 0.09999999999999999, '#666666'],
[4.0, 2.0, 0.05, '#666666'],
[6.0, 4.0, 0.12142857142857143, '#666666'],
[10.0, 2.0, 0.07857142857142858, '#666666']])]
self._matchWeightedData(match, target)
if show is True:
p = graph.plot.Dolan(s, title='Dynamics')
p.run()
def testPartReductionC(self):
'''
Artificially create test cases.
'''
from music21 import analysis
from music21 import dynamics
s = stream.Score()
p1 = stream.Part()
p1.id = 0
p2 = stream.Part()
p2.id = 1
for ql in [1, 2, 1, 4]:
p1.append(note.Note(quarterLength=ql))
p2.append(note.Note(quarterLength=ql))
for pos, dyn in [(0, 'p'), (2, 'fff'), (6, 'ppp')]:
p1.insert(pos, dynamics.Dynamic(dyn))
for pos, dyn in [(0, 'p'), (1, 'fff'), (2, 'ppp')]:
p2.insert(pos, dynamics.Dynamic(dyn))
s.insert(0, p1)
s.insert(0, p2)
# s.show()
pr = analysis.reduction.PartReduction(s, normalize=False)
pr.process()
match = pr.getGraphHorizontalBarWeightedData()
target = [(0, [[0.0, 2.0, 0.05, '#666666'],
[2.0, 4.0, 0.1285714285714286, '#666666'],
[6.0, 2.0, 0.0214285714286, '#666666']]),
(1, [[0.0, 1.0, 0.05, '#666666'],
[1.0, 1.0, 0.1285714285714286, '#666666'],
[2.0, 6.0, 0.0214285714286, '#666666']])]
self._matchWeightedData(match, target)
def testPartReductionD(self):
'''
Artificially create test cases. Here, uses rests.
'''
from music21 import analysis
from music21 import dynamics
s = stream.Score()
p1 = stream.Part()
p1.id = 0
p2 = stream.Part()
p2.id = 1
for ql in [None, 2, False, 2, False, 2]:
if ql:
p1.append(note.Note(quarterLength=ql))
p2.append(note.Note(quarterLength=ql))
else:
p1.append(note.Rest(quarterLength=2))
p2.append(note.Rest(quarterLength=2))
for pos, dyn in [(0, 'p'), (2, 'fff'), (6, 'ppp')]:
p1.insert(pos, dynamics.Dynamic(dyn))
for pos, dyn in [(0, 'mf'), (2, 'f'), (6, 'mf')]:
p2.insert(pos, dynamics.Dynamic(dyn))
s.insert(0, p1)
s.insert(0, p2)
# s.show()
pr = analysis.reduction.PartReduction(s)
pr.process()
match = pr.getGraphHorizontalBarWeightedData()
# print(match)
target = [(0, [[2.0, 2.0, 1.0, '#666666'],
[6.0, 2.0, 1 / 6, '#666666'],
[10.0, 2.0, 1 / 6, '#666666']]),
(1, [[2.0, 2.0, 7 / 9, '#666666'],
[6.0, 2.0, 0.6111111111111112, '#666666'],
[10.0, 2.0, 0.6111111111111112, '#666666']])]
self._matchWeightedData(match, target)
# p = graph.PlotDolan(s, title='Dynamics')
# p.process()
def testPartReductionE(self):
'''
Artificially create test cases.
'''
from music21 import analysis
from music21 import dynamics
s = stream.Score()
p1 = stream.Part()
p1.id = 0
p2 = stream.Part()
p2.id = 1
for ql in [2, 2, False, 2, False, 2]:
if ql:
p1.append(note.Note(quarterLength=ql))
p2.append(note.Note(quarterLength=ql))
else:
p1.append(note.Rest(quarterLength=2))
p2.append(note.Rest(quarterLength=2))
for pos, dyn in [(0, 'p'), (2, 'fff'), (6, 'ppp')]:
p1.insert(pos, dynamics.Dynamic(dyn))
for pos, dyn in [(0, 'mf'), (2, 'f'), (6, 'mf')]:
p2.insert(pos, dynamics.Dynamic(dyn))
p1.makeMeasures(inPlace=True)
p2.makeMeasures(inPlace=True)
s.insert(0, p1)
s.insert(0, p2)
# s.show()
pr = analysis.reduction.PartReduction(s, fillByMeasure=True,
segmentByTarget=False, normalize=False)
pr.process()
target = pr.getGraphHorizontalBarWeightedData()
match = [(0, [[0.0, 4.0, 0.178571428571, '#666666'],
[4.0, 4.0, 0.0214285714286, '#666666'],
[8.0, 4.0, 0.0214285714286, '#666666']]),
(1, [[0.0, 4.0, 0.178571428571, '#666666'],
[4.0, 4.0, 0.07857142857142858, '#666666'],
[8.0, 4.0, 0.07857142857142858, '#666666']])]
self._matchWeightedData(match, target)
pr = analysis.reduction.PartReduction(s, fillByMeasure=False,
segmentByTarget=True, normalize=False)
pr.process()
target = pr.getGraphHorizontalBarWeightedData()
match = [(0, [[0.0, 2.0, 0.05, '#666666'],
[2.0, 2.0, 0.1285714285714286, '#666666'],
[6.0, 2.0, 0.0214285714286, '#666666'],
[10.0, 2.0, 0.0214285714286, '#666666']]),
(1, [[0.0, 2.0, 0.07857142857142858, '#666666'],
[2.0, 2.0, 0.1, '#666666'],
[6.0, 2.0, 0.07857142857142858, '#666666'],
[10.0, 2.0, 0.07857142857142858, '#666666']])]
# from pprint import pprint as print
# print(target)
self._matchWeightedData(match, target)
pr = analysis.reduction.PartReduction(s, fillByMeasure=False,
segmentByTarget=False)
pr.process()
target = pr.getGraphHorizontalBarWeightedData()
# print(target)
match = [(0, [[0.0, 4.0, 1.0, '#666666'],
[6.0, 2.0, 0.12, '#666666'],
[10.0, 2.0, 0.12, '#666666']]),
(1, [[0.0, 4.0, 1.0, '#666666'],
[6.0, 2.0, 0.44, '#666666'],
[10.0, 2.0, 0.44, '#666666']])]
self._matchWeightedData(match, target)
pr = analysis.reduction.PartReduction(s, fillByMeasure=True,
segmentByTarget=True)
pr.process()
target = pr.getGraphHorizontalBarWeightedData()
match = [(0, [[0.0, 2.0, 0.3888888888888, '#666666'],
[2.0, 2.0, 1.0, '#666666'],
[6.0, 2.0, 0.166666666667, '#666666'],
[8.0, 4.0, 0.166666666667, '#666666']]),
(1, [[0.0, 2.0, 0.6111111111111112, '#666666'],
[2.0, 2.0, 0.7777777777777776, '#666666'],
[6.0, 2.0, 0.611111111111111, '#666666'],
[8.0, 4.0, 0.611111111111111, '#666666']])]
self._matchWeightedData(match, target)
# p = graph.PlotDolan(s, title='Dynamics', fillByMeasure=False,
# segmentByTarget=True, normalizeByPart=False)
# p.process()
def xtestPartReductionSchoenberg(self):
from music21 import corpus
sc = corpus.parse('schoenberg/opus19', 2)
pr = PartReduction(
sc,
fillByMeasure=False,
segmentByTarget=True,
normalizeByPart=False
)
pr.process()
unused_target = pr.getGraphHorizontalBarWeightedData()
class TestExternal(unittest.TestCase):
show = True
def testPartReductionB(self):
test = Test()
test.testPartReductionB(show=self.show)
# ------------------------------------------------------------------------------
# define presented order in documentation
_DOC_ORDER: DocOrder = []
if __name__ == '__main__':
import music21
music21.mainTest(Test) # , runTest='testPartReductionSchoenberg')