diff Functions/MeasureFunctions.py @ 1:0f7f611deca4

Functions

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/Functions/MeasureFunctions.py	Mon May 04 22:53:31 2015 +0200
@@ -0,0 +1,527 @@
+'''
+@organization: Lancaster University & University of Leeds
+@version: 1.0
+Created on 11/12/2014
+
+@author: Victor Padilla
+@contact: [e-mail]
+
+Functions to manipulate measures
+
+
+'''
+
+from music21 import stream
+from music21 import note
+from music21 import meter
+from music21 import omr
+from music21 import spanner
+from music21 import chord
+import copy
+
+
+class MeasureFunctions:
+    
+    def getDuration(self,bar):
+        '''
+        Returns the duration of a single measure
+        '''
+        duration=0
+        try:
+            for event in bar:
+                try:
+                    if(event.isNote):
+                        duration+=event.duration.quarterLength
+                    if(event.isRest):
+                        duration+=event.duration.quarterLength
+                except:
+                    pass
+        except:
+            pass
+        return str(duration)
+    
+    def correctIncorrectMeasuresArray(self,omr,incorrectMeasures):
+        '''
+        Iteractive method that detects if some measures have been wrong flagged as an error.
+        It is under development
+        '''
+        measures=omr.parts[0].getElementsByClass(stream.Measure)
+        if 0 in incorrectMeasures:
+            incorrectMeasures.remove(0)#Anacrusis
+            
+        for barNumber in incorrectMeasures:
+            if barNumber<len(measures)-1:
+                measure=measures[barNumber]
+                measureNext=measures[barNumber+1]
+                duration=self.getDuration(measure)
+                
+                if(float(duration)<=0):
+                    incorrectMeasures.remove(barNumber)
+                    self.correctIncorrectMeasuresArray(omr,incorrectMeasures)
+                    
+                if(measureNext!=None):
+                    durationNext=self.getDuration(measureNext)
+                    if(float(duration)+float(durationNext)==4):
+                        try:
+                            incorrectMeasures.remove(barNumber)
+                            incorrectMeasures.remove(barNumber+1)
+                        except:
+                            pass
+                        self.correctIncorrectMeasuresArray(omr,incorrectMeasures)
+        return incorrectMeasures
+    
+    
+    def filterExtraMeasures(self,omr):
+        '''
+        Function that removes empty measures
+        '''
+        sco=stream.Score()
+        s=stream.Part()
+        for measure in omr.parts[0].getElementsByClass(stream.Measure):
+            if measure.flat.duration.quarterLength>0:
+                s.append(measure)
+        sco.append(s)
+        return sco
+    
+    
+    def getIncorrectMeasureIndices(self,omr):
+        '''
+        Method that detect wrong measures using transitions
+        '''
+        arrChunks= self.getTransitions(omr)
+        measures=omr.parts[0].getElementsByClass(stream.Measure)
+        indexBar=0
+        FlagErrors=[]
+        barFrom=0
+        barTo=0
+        for chunk in arrChunks:
+            indexC=arrChunks.index(chunk)
+            chunkBefore=arrChunks[indexC-1]
+            if(indexC==0):
+                barFrom=0
+            else:
+                barFrom+=chunkBefore[1]  
+            barTo=chunk[1]+barFrom       
+            chunkMeasures=measures[barFrom:barTo]
+            quarterChunk=round(chunk[0]/2)
+            for measure in chunkMeasures:              
+                if measure.duration.quarterLength!=quarterChunk:
+                    FlagErrors.append(indexBar)
+                indexBar+=1
+        return FlagErrors
+    
+    def _filterTransitions(self,arrMeasureIndex):
+        '''
+        Inner method that removes false transitions in measures for detecting missing time signatures
+        '''
+        arrMeasureIndex2=[]
+        arrMeasureIndex.insert(0,0)
+        arrMeasureIndex.append(-1) 
+        for mes in arrMeasureIndex:
+            indexM=arrMeasureIndex.index(mes)
+            if indexM>0:
+                bars=arrMeasureIndex[indexM]-arrMeasureIndex[indexM-1]
+                if (bars>9):
+                    arrMeasureIndex2.append(mes)
+        arrMeasureIndex2.insert(0,0)
+        arrMeasureIndex2.append(-1) 
+        return arrMeasureIndex2
+   
+    def getTransitions(self,omr):
+        '''
+        Returns transitions in a omr
+        '''
+        MeasuresLength=[]
+        arrMeasureIndex_before=self._getTransitionBar(omr)
+        
+        arrMeasureIndex=self._filterTransitions(arrMeasureIndex_before)
+        for i in range(len(arrMeasureIndex)-1):
+            arrMeasureslength= self._getAverageQuavers(omr,arrMeasureIndex[i],arrMeasureIndex[i+1],False)
+            MeasuresLength.append(arrMeasureslength)
+
+        return MeasuresLength
+        
+    def _getTransitionBar(self,omr):
+        '''
+        Inner function for getting transitions
+        '''
+        arrOut=[]
+        measures=omr.parts[0].getElementsByClass(stream.Measure)
+        barsNumber=len(measures)
+        for barIndex in range(5,barsNumber):
+            averageBefore=self._getAverageQuavers(omr,barIndex-4,barIndex,isUntilTS=False)
+            averageAfter=self._getAverageQuavers(omr,barIndex,barIndex+2,isUntilTS=False)
+            if(abs(averageBefore[0]-averageAfter[0])>1.5):
+                if(measures[barIndex].duration.quarterLength<averageBefore[0]*2):#rest bars or missing bars
+                    arrOut.append(barIndex)
+        return arrOut
+        
+    def _getAverageQuavers(self,myStream,measureIndex,measureIndexEnd,isUntilTS):
+        '''
+        Inner function for getting the average quavers per measure in a group of measures
+        '''
+        quavers=0
+        barNumbers=0
+        averageQuavers=0
+        measures=myStream.parts[0].getElementsByClass(stream.Measure)
+        for bar in measures[measureIndex:measureIndexEnd]:
+            duration=bar.duration.quarterLength*2
+            barNumbers+=1
+            if(duration>0 and duration<10):
+                quavers+=duration
+                if isUntilTS:
+                    if (len(bar.getElementsByClass(meter.TimeSignature))>0):
+                        break
+        if barNumbers>0:   
+            averageQuavers=quavers/barNumbers
+        return averageQuavers,barNumbers
+    
+#**************************************************
+    def getPossibleBeamsErrors(self,omr):   
+        '''
+        Function that returns measure errors based on non conventional beaming
+        '''
+        arrErrors=[] 
+        measures=omr.parts[0].getElementsByClass(stream.Measure)
+        barsNumber=len(measures)
+        for i in range(barsNumber):
+            notes=measures[i].getElementsByClass(note.Note)
+            count=0
+            state=0
+            for n in notes:
+                if n.duration.quarterLength==0.25:
+                    bs=n.beams.getNumbers() 
+                    if(len(bs)>0):
+                        b=n.beams.getByNumber(bs[0])          
+                        if b.type=='start':
+                            count=1
+                            state=1
+                        if b.type=='continue':
+                            if(state==1 or state==2):
+                                count+=1
+                                state=2
+                        if b.type=='stop':
+                            if(state==1 or state==2):
+                                count+=1
+                                if count==3:
+                                    arrErrors.append(i)
+        arrErrors=list(set(arrErrors))
+        return arrErrors
+
+    def getPossibleLastNoteErrors(self,omr):   
+        '''
+        Function that returns measure errors based on the last notes rhythm
+        '''
+        arrErrors=[] 
+        measures=omr.parts[0].getElementsByClass(stream.Measure)
+        barsNumber=len(measures)
+        for i in range(barsNumber):
+            notes=measures[i].getElementsByClass(note.GeneralNote)
+            if(len(notes)>0):
+                lastNote=notes[len(notes)-1]
+                if lastNote.duration.quarterLength<=0.25:
+                    if(lastNote.isRest):
+                        arrErrors.append(i)
+                    else:
+                        bs=lastNote.beams.getNumbers() 
+                        
+                        if(len(bs)==0):
+                            if(len(notes)>2):
+                                noteBefore=notes[len(notes)-2]
+                                if noteBefore.duration.quarterLength+lastNote.duration.quarterLength!=1:
+                                    arrErrors.append(i)
+               
+            
+        return arrErrors
+    
+    def flagIncorrectMeasures(self,omr2):
+        '''
+        flag the incorrect measures from a omr using different methods
+        '''
+        mf=MeasureFunctions()
+        sc=omr.correctors.ScoreCorrector(omr2)
+        part=sc.getSinglePart(0)
+         
+        arrErrors=[]
+        im=part.getIncorrectMeasureIndices(runFast=False)
+        
+        im1= mf.getIncorrectMeasureIndices(omr2)
+        
+        im2= mf.getPossibleBeamsErrors(omr2)
+        im3= mf.getPossibleLastNoteErrors(omr2)
+       
+         
+        arrErrors.append(im)
+        arrErrors.append(im1)
+        arrErrors.append(im2)
+        arrErrors.append(im3)
+         
+        if(len(im)>15):
+            if(len(im1)<len(im)):
+                im=im1
+        imSum=list(set(im)|set(im2)|set(im3))
+        imSum=sorted(imSum)
+        imOK=mf.correctIncorrectMeasuresArray(omr2,imSum)  
+        print im
+        return imOK,arrErrors
+    
+    def getSlurs(self,part):
+        '''
+        Returns the slurs from one part
+        '''
+        slurs=part.flat.getElementsByClass(spanner.Spanner)
+        return slurs
+        
+    def reconstructScore(self,part,hashPart): 
+        '''
+        This function include rest measures in one part based on the hash part aligned
+        '''
+        partReconstructed=stream.Part()
+        barNumber=1
+        gaps=[]
+        for i in range(len(hashPart)): 
+            if hashPart[i]!="*":
+                m=part.getElementsByClass(stream.Measure)[barNumber-1]
+                partReconstructed.append(m)
+                barNumber+=1
+            else:
+                m=stream.Measure()
+                partReconstructed.append(m)
+                gaps.append(i)
+        slurs=self.getSlurs(part)   
+        partReconstructed.append(slurs)  
+        myStream=self.reorderMeasures(partReconstructed)  
+        return myStream,gaps
+    
+    def reorderMeasures(self,omr):
+        '''
+        Returns the correct measures number
+        '''
+        slurs=self.getSlurs(omr) 
+        s=stream.Part()
+        barNumber=1
+        for measure in omr.getElementsByClass(stream.Measure):
+            measure.number=barNumber
+            s.append(measure)
+            barNumber+=1
+        s.append(slurs)
+        return s      
+    
+    def getNoteByOffset(self,voice,offset):
+        '''
+        Returns one note in one voice according to the offset
+        '''
+        for n in voice.getElementsByClass(note.Note):
+            if n.offset==offset:
+                return n
+        for n in voice.getElementsByClass(chord.Chord):
+            if n.offset==offset:
+                return n
+    
+    def convertVoicesToChord(self,omr):
+        '''
+        Function that converts voices with the same rhythm to chords
+        '''
+        for part in omr.getElementsByClass(stream.Part):
+            for measure in part.getElementsByClass(stream.Measure):
+                measure.show('text')
+                voices=measure.getElementsByClass(stream.Voice)
+                if len(voices)>=2:
+                    for element1 in voices[0].getElementsByClass(note.GeneralNote):
+                        offset1= element1.offset
+                        for voice in voices:
+                            if voice!=voices[0]:
+                                element2=self.getNoteByOffset( voice, offset1)
+                                if element2 is not None and element1.duration.quarterLength==element2.duration.quarterLength and not isinstance(element1,note.Rest):
+                                    mychord=self.mergeChords(element1,element2)  
+                                    mychord.duration.quarterLength=element1.quarterLength
+                                    voices[0].replace(element1,mychord)
+                                    myrest=note.Rest()
+                                    myrest.duration.quarterLength=element1.quarterLength
+                                    voice.replace(element2,myrest)
+
+        omr=self.removeRestVoice(omr)
+        return omr
+    def removeRestVoice(self,omr):
+        '''
+        This option removes the rest voices (without notes) 
+        '''
+        for part in omr.getElementsByClass(stream.Part):
+            for measure in part.getElementsByClass(stream.Measure):
+                voices=measure.getElementsByClass(stream.Voice)
+                for voice in voices:
+                    myvoice = copy.deepcopy(voice)
+                    myvoice.removeByClass('Rest')
+                    if len(myvoice)==0:
+                        measure.remove(voice)
+        
+        return omr
+     
+    def mergeChords(self,element1,element2): 
+        '''
+        Returns one chord made by two elements, notes or chords
+        '''
+        notes1=[]
+        notes2=[]
+        notes=[]
+        if isinstance(element1,chord.Chord):
+            notes1=element1.pitches
+        else:
+            notes1.append(element1.pitch)
+        
+        if isinstance(element2,chord.Chord):
+            notes2=element2.pitches
+        else:
+            notes2.append(element2.pitch)
+        
+        for n in notes1:
+            notes.append(n)
+        for n in notes2:
+            notes.append(n)
+
+        mychord = chord.Chord(notes)
+        return mychord
+        
+    
+    def convertBeamsToTriplets(self,omr):  
+        '''
+        This option search for measures where the rhythm is bigger than the time signature
+        and tries to convert 3 note beams in triplets
+        '''
+        tsDefault=omr.flat.getElementsByClass('TimeSignature')[0]
+        idPart=0
+        for part in omr.getElementsByClass(stream.Part):
+            idPart+=1
+            idMeasure=0
+            for measure in part.getElementsByClass(stream.Measure):
+                idMeasure+=1
+                if self._IsMeasureHigherTS(idMeasure, part,tsDefault):
+                    print idMeasure
+                    measure=self.beamsToTriplets(measure)
+        return omr
+               
+                    
+    def _IsMeasureHigherTS(self,idMeasure,part,ts):
+        '''
+        Inner function that detects if one measure is bigger than the time signature
+        '''
+        idCountMeasure=0
+        for measure in part.getElementsByClass(stream.Measure):
+            idCountMeasure+=1
+            newts=measure.flat.getElementsByClass('TimeSignature')
+            if len(newts)>0:
+                ts=newts[0]
+            if idMeasure==idCountMeasure:
+                voices=measure.getElementsByClass(stream.Voice)
+                totalDuration=0
+                if len(voices)>=1:
+                    for voice in voices:
+                        totalDuration=0
+                        for element in voice.getElementsByClass(note.GeneralNote):
+                            totalDuration+=element.duration.quarterLength;
+                else:
+                    for element in measure.getElementsByClass(note.GeneralNote):
+                        totalDuration+=element.duration.quarterLength;
+                quartersBymeasure=ts.numerator*(4.0/ts.denominator)
+                if totalDuration>quartersBymeasure:
+                    return True
+                else:
+                    return False
+    def beamsToTriplets(self,measure):
+        '''
+        This function returns one measure changing 3 notes beamed by triplets
+        '''
+        voices=measure.getElementsByClass(stream.Voice)
+        if len(voices)>=1:
+            for voice in voices:
+                notes=voice.getElementsByClass(note.NotRest)
+                notes=self._notesToTriplets(notes)
+               
+        else:
+            notes=measure.getElementsByClass(note.NotRest)
+            notes=self._notesToTriplets(notes)
+                
+        return measure
+    
+    
+    def _notesToTriplets(self,notes):
+        '''
+        Gets one note arrays and tries to convert to triplets
+        '''
+
+        for i in range(len(notes)):
+            self._removeSlurs(notes[i])
+            beams1=notes[i].beams.beamsList
+            
+            try:
+                beams2=notes[i+1].beams.beamsList
+                beams3=notes[i+2].beams.beamsList
+                if len(beams1)>0 and len(beams2)>0 and len(beams3)>0:
+                    if beams1[0].type=='start' and beams2[0].type=='continue' and beams3[0].type=='stop':
+                        if notes[i].duration.quarterLength==notes[i+1].duration.quarterLength==notes[i+2].duration.quarterLength:
+                            mytype=notes[i].duration.type
+                            duration=note.duration.convertTypeToQuarterLength(mytype)
+                            realDuration=duration*2.0/3.0
+                            notes[i].duration.quarterLength=realDuration
+                            notes[i+1].duration.quarterLength=realDuration
+                            notes[i+2].duration.quarterLength=realDuration
+            
+            except:
+                pass
+        
+        return notes
+                            
+    def _removeSlurs(self,n): 
+        '''
+        Removes the slur in one note
+        '''
+        for element in n.getSites():
+            if isinstance(element,stream.SpannerStorage):
+                for e in element.elements:
+                    element.pop(0)   
+    
+    def removesEmptyVoices(self,omr):
+        '''
+        Removes empty voices (just rests)
+        '''
+        for part in omr.getElementsByClass(stream.Part):
+            for measure in part.getElementsByClass(stream.Measure):
+                print measure
+                voices=measure.getElementsByClass(stream.Voice)
+                if len(voices)>=1:
+                    print len(voices)
+                    for voice in voices:
+                        notes=voice.getElementsByClass(note.NotRest)
+                        if len(notes)==0:
+                            index=measure.index(voice)
+                            measure.pop(index)
+                voices=measure.getElementsByClass(stream.Voice)
+                if len(voices)==1:
+                    measure.flattenUnnecessaryVoices()
+        return omr
+    
+    def removesGaps(self,omr):
+        '''
+        This option changes gaps by rests in voices
+        '''
+        for part in omr.getElementsByClass(stream.Part):
+            for measure in part.getElementsByClass(stream.Measure):
+                voices=measure.getElementsByClass(stream.Voice)
+                if len(voices)>=1:
+                    for voice in voices:
+                        v=voice.findGaps()
+                        if v!=None:
+                            nextOffset=1000
+                            for element in voice.getElementsByClass(note.GeneralNote):
+                                offset= element.offset
+                                if offset>nextOffset:
+                                    rest=note.Rest()
+                                    rest.duration.quarterLength=offset-nextOffset
+                                    voice.insert(nextOffset,rest)
+                                    nextOffset=1000
+                                else:
+                                    duration=element.duration.quarterLength
+                                    nextOffset=offset+duration
+        return omr                    
+                        
+