# -*- coding: utf-8 -*- # ------------------------------------------------------------------------------ # Name: testPerformance.py # Purpose: Tests keep track of long-term performance targets # # Authors: Christopher Ariza # # Copyright: Copyright © 2009-2011 Michael Scott Asato Cuthbert # License: BSD, see license.txt # ------------------------------------------------------------------------------ ''' This module defines a number of performance test. Results for these performances are stored and dated, and used to track long-term performance changes. This file is not run with the standard test battery presently. ''' from __future__ import annotations import unittest import music21 from music21 import common from music21 import corpus from music21 import environment from music21.musicxml.m21ToXml import GeneralObjectExporter as GEX environLocal = environment.Environment('test.testPerformance') # ------------------------------------------------------------------------------ class Test(unittest.TestCase): def runStreamIterationByIterator(self): ''' Stream iteration by iterator ''' from music21 import note from music21 import stream # create a stream with 750 notes, 250 rests s = stream.Stream() for i in range(1000): n = note.Note() s.append(n) for i in range(500): r = note.Rest() s.append(r) for i in range(100): for j in s: # this will create an iterator instances pass def runStreamIterationByElements(self): ''' Stream iteration by .elements access ''' from music21 import note from music21 import stream # create a stream with 750 notes, 250 rests s = stream.Stream() for i in range(1000): n = note.Note() s.append(n) for i in range(500): r = note.Rest() s.append(r) for i in range(100): for j in s.elements: # this will create an iterator instances pass def runGetElementsByClassType(self): ''' Getting elements by class type ''' from music21 import note from music21 import stream # create a stream with 750 notes, 250 rests s = stream.Stream() for i in range(1000): n = note.Note() s.append(n) for i in range(500): r = note.Rest() s.append(r) for i in range(2): post = s.recurse().getElementsByClass([note.Rest, note.Note]) self.assertEqual(len(post), 1500) def runGetElementsByClassString(self): ''' Getting elements by string ''' from music21 import note from music21 import stream # create a stream with 750 notes, 250 rests s = stream.Stream() for i in range(1000): n = note.Note() s.append(n) for i in range(500): r = note.Rest() s.append(r) for i in range(2): post = s.recurse().getElementsByClass(['Rest', 'Note']) self.assertEqual(len(post), 1500) def runParseBeethoven(self): ''' Loading file: beethoven/opus59no2/movement3 ''' junk = corpus.parse('beethoven/opus59no2/movement3', forceSource=True) def runMusicxmlOutPartsBeethoven(self): ''' Loading file and rendering musicxml output for each part: beethoven/opus59no2/movement3 ''' x = corpus.parse('beethoven/opus59no2/movement3', forceSource=True) # problem: doing each part is much faster than the whole score for p in x.parts: junk = GEX().parse(p) def runMusicxmlOutScoreBeethoven(self): ''' Loading file and rendering musicxml output of complete score: beethoven/opus59no2/movement3 ''' x = corpus.parse('beethoven/opus59no2/movement3', forceSource=True) # problem: doing each part is much faster than the whole score junk = GEX().parse(x) def runParseHaydn(self): ''' Loading file: haydn/opus74no1/movement3 ''' junk = corpus.parse('haydn/opus74no1/movement3', forceSource=True) def runParseRobertSchumann(self): ''' Loading file: schumann_robert/opus41no1/movement2 ''' junk = corpus.parse('schumann_robert/opus41no1/movement2', forceSource=True) def runParseLuca(self): ''' Loading file: luca/gloria ''' junk = corpus.parse('luca/gloria', forceSource=True) def runMusicxmlOutLuca(self): ''' Loading file and rendering musicxml output: luca/gloria ''' x = corpus.parse('luca/gloria', forceSource=True) junk = GEX().parse(x) def runCreateTimeSignatures(self): ''' Creating 500 TimeSignature objects ''' from music21 import meter tsStr = ['4/4', '4/4', '4/4', '3/4', '3/4', '2/4', '2/4', '2/2', '2/2', '3/8', '6/8', '9/8', '5/4', '12/8'] for i in range(500): meter.TimeSignature(tsStr[i % len(tsStr)]) def runCreateDurations(self): ''' Creating 10000 Duration objects ''' from music21 import duration qlList = [4, 2, 1, 0.5, 1 / 3, 0.25, 0.125] for i in range(10000): ql = qlList[i % len(qlList)] d = duration.Duration() d.quarterLength = ql junk = d.quarterLength def runCreatePitches(self): ''' Creating 50000 Pitch objects ''' from music21 import pitch pList = [1.5, 5, 20.333333, 8, 2.5, 'A#', 'b`', 'c6#~'] for i in range(50000): inputPName = pList[i % len(pList)] p = pitch.Pitch(inputPName) p.transpose('p5', inPlace=True) def runParseABC(self): ''' Creating loading a large multiple work abc file (han1) ''' dummy = corpus.parse('essenFolksong/han1') def runGetElementsByContext(self): ''' Test getting elements by context from a Stream ''' from music21 import stream from music21 import clef from music21 import key from music21 import meter s = corpus.parse('bwv66.6') # create a few secondary streams to add more sites unused_flat = s.flatten() unused_notes = s.flatten().notes for p in s.parts: for m in p.getElementsByClass(stream.Measure): if m.number == 0: continue post = m.getContextByClass(clef.Clef) assert post is not None post = m.getContextByClass(meter.TimeSignature) assert post is not None post = m.getContextByClass(key.KeySignature) assert post is not None for n in m.notesAndRests: post = n.getContextByClass(clef.Clef) assert post is not None post = n.getContextByClass(meter.TimeSignature) assert post is not None post = n.getContextByClass(key.KeySignature) assert post is not None def runGetElementsByPrevious(self): ''' Test getting elements by using the previous method ''' from music21 import stream from music21 import clef from music21 import key from music21 import meter s = corpus.parse('bwv66.6') # create a few secondary streams to add more sites unused_flat = s.flatten() unused_notes = s.flatten().notes for p in s.parts: for m in p.getElementsByClass(stream.Measure): if m.number == 0: continue post = m.previous(clef.Clef) assert post is not None post = m.previous(meter.TimeSignature) assert post is not None post = m.previous(key.KeySignature) assert post is not None for n in m.notesAndRests: post = n.getContextByClass(clef.Clef) assert post is not None post = n.getContextByClass(meter.TimeSignature) assert post is not None post = n.getContextByClass(key.KeySignature) assert post is not None def runParseMonteverdiRNText(self): ''' Loading file: beethoven/opus59no2/movement3 ''' unused = corpus.parse('monteverdi/madrigal.5.3.rntxt', forceSource=True) # -------------------------------------------------------------------------- def testTimingTolerance(self): ''' Test the performance of methods defined above, comparing the resulting time to the time obtained in past runs. This should not produce errors as such, but is used to provide reference if overall performance has changed. ''' # provide work and expected min/max in seconds for testMethod, best in [ (self.runGetElementsByPrevious, { '2011.11.29': 4.69, }), (self.runGetElementsByContext, { '2010.11.10': 7.3888170, '2010.11.11': 3.96121883392, }), # # # # # (self.runParseABC, # # { # # '2010.10.20': 38.6760668755, # # '2010.10.21': 35.4297668934, # # }), # # # (self.runCreateDurations, # { # '2010.10.07': 0.201117992401, # # }), # # (self.runCreateTimeSignatures, # { # '2010.10.07': 2.88308691978, # '2010.10.08': 1.40892004967 , # # }), # # # (self.runStreamIterationByIterator, # {'2010.09.20': 2.2524, # '2010.10.07': 1.8214, # }), # # (self.runStreamIterationByElements, # {'2010.09.20': 0.8317, # }), # # # (self.runGetElementsByClassType, # {'2010.09.20': 3.28, # }), # # (self.runGetElementsByClassString, # {'2010.09.20': 3.22, # }), # # (self.runParseBeethoven, # {'2009.12.14': 7.42, # '2009.12.15': 6.686, # '2010.06.24': 7.475, # '2010.07.08': 3.562, # }), # # (self.runMusicxmlOutPartsBeethoven, # {'2010.09.20': 7.706, # }), # # (self.runMusicxmlOutScoreBeethoven, # {'2010.09.20': 33.273, # '2010.10.07': 11.9290, # }), # # (self.runCreatePitches, # {'2011.04.12': 31.071, # }), # (self.runParseHaydn, # {'2009.12.14': 4.08, # '2009.12.15': 3.531, # '2010.06.24': 3.932, # '2010.07.08': 1.935, # }), # (self.runParseRobertSchumann, # {'2009.12.14': 5.88, # '2009.12.15': 5.126, # '2010.06.24': 5.799, # '2010.07.08': 2.761, # }), # (self.runParseLuca, # {'2009.12.14': 3.174, # '2009.12.15': 2.954, # '2010.06.24': 3.063, # '2010.07.08': 1.508, # }), # # (self.runMusicxmlOutLuca, # {'2010.09.20': 8.372, # '2010.10.07': 4.5613, # }), # # (self.runParseMonteverdiRNText, # {'2011.02.27': 6.411, # '2011.02.28': 2.944, # }), ]: # end of long for loop t = common.Timer() t.start() # call the test testMethod() t.stop() dur = t() items = list(best.items()) items.sort() items.reverse() environLocal.printDebug(['\n\ntiming tolerance for:', str(testMethod.__doc__.strip()), '\nthis run:', dur, '\nbest runs:', [f'{x}: {y}' for x, y in items], '\n' ] ) # self.assertEqual(True, dur <= max) # performance test if __name__ == '__main__': import sys if len(sys.argv) == 1: # normal conditions # sys.arg test options will be used in mainTest() music21.mainTest(Test)