]> git.donarmstrong.com Git - lilypond.git/blobdiff - scripts/musicxml2ly.py
Fix some bugs in the dynamic engraver and PostScript backend
[lilypond.git] / scripts / musicxml2ly.py
index b132176390250843f927874a63529b976d7398f6..5f25567c592909beb52fca77bedffcb433fd9cbc 100644 (file)
@@ -1,4 +1,4 @@
-#!@PYTHON@
+#!@TARGET_PYTHON@
 
 import optparse
 import sys
@@ -7,396 +7,483 @@ import os
 import string
 from gettext import gettext as _
 
+
+
+
 datadir = '@local_lilypond_datadir@'
 if not os.path.isdir (datadir):
-       datadir = '@lilypond_datadir@'
+    datadir = '@lilypond_datadir@'
 if os.environ.has_key ('LILYPONDPREFIX'):
-       datadir = os.environ['LILYPONDPREFIX']
-       while datadir[-1] == os.sep:
-               datadir = datadir[:-1]
+    datadir = os.environ['LILYPONDPREFIX']
+    while datadir[-1] == os.sep:
+        datadir = datadir[:-1]
 
 if os.path.exists (os.path.join (datadir, 'share/lilypond/@TOPLEVEL_VERSION@/')):
-       datadir = os.path.join (datadir, 'share/lilypond/@TOPLEVEL_VERSION@/')
+    datadir = os.path.join (datadir, 'share/lilypond/@TOPLEVEL_VERSION@/')
 elif os.path.exists (os.path.join (datadir, 'share/lilypond/current/')):
-       datadir = os.path.join (datadir, 'share/lilypond/current/')
+    datadir = os.path.join (datadir, 'share/lilypond/current/')
 
 sys.path.insert (0, os.path.join (datadir, 'python'))
 
+# dynamic relocation, for GUB binaries.
+bindir = os.path.split (sys.argv[0])[0]
+
+for prefix_component in ['share', 'lib']:
+    datadir = os.path.abspath (bindir + '/../%s/lilypond/current/python/' % prefix_component)
+    sys.path.insert (0, datadir)
+
+
+import lilylib as ly
 
 import musicxml
 import musicexp
+
 from rational import Rational
 
 
 def progress (str):
-       sys.stderr.write (str + '\n')
-       sys.stderr.flush ()
-       
+    sys.stderr.write (str + '\n')
+    sys.stderr.flush ()
+    
 
 def musicxml_duration_to_lily (mxl_note):
-       d = musicexp.Duration ()
-       if mxl_note.get_maybe_exist_typed_child (musicxml.Type):
-               d.duration_log = mxl_note.get_duration_log ()
-       else:
-               d.duration_log = 0
+    d = musicexp.Duration ()
+    if mxl_note.get_maybe_exist_typed_child (musicxml.Type):
+        d.duration_log = mxl_note.get_duration_log ()
+    else:
+        d.duration_log = 0
 
-       d.dots = len (mxl_note.get_typed_children (musicxml.Dot))
-       d.factor = mxl_note._duration / d.get_length ()
+    d.dots = len (mxl_note.get_typed_children (musicxml.Dot))
+    d.factor = mxl_note._duration / d.get_length ()
 
-       return d        
+    return d         
 
 def group_tuplets (music_list, events):
 
 
-       """Collect Musics from
-       MUSIC_LIST demarcated by EVENTS_LIST in TimeScaledMusic objects.
-       """
-
-       
-       indices = []
-
-       j = 0
-       for (ev_chord, tuplet_elt, fraction) in events:
-               while (j < len (music_list)):
-                       if music_list[j]== ev_chord:
-                               break
-                       j += 1
-               if tuplet_elt.type == 'start':
-                       indices.append ((j, None, fraction))
-               elif tuplet_elt.type == 'stop':
-                       indices[-1] = (indices[-1][0], j, indices[-1][2])
-
-       new_list = []
-       last = 0
-       for (i1, i2, frac) in indices:
-               if i1 >= i2:
-                       continue
-
-               new_list.extend (music_list[last:i1])
-               seq = musicexp.SequentialMusic ()
-               last = i2 + 1
-               seq.elements = music_list[i1:last]
-
-               tsm = musicexp.TimeScaledMusic ()
-               tsm.element = seq
-
-               tsm.numerator = frac[0]
-               tsm.denominator  = frac[1]
-
-               new_list.append (tsm)
-
-       new_list.extend (music_list[last:])
-       return new_list
-
-
-def musicxml_clef_to_lily (mxl):
-       sign = mxl.get_maybe_exist_named_child ('sign')
-       change = musicexp.ClefChange ()
-       if sign:
-               change.type = sign.get_text ()
-       return change
-
-       
-def musicxml_time_to_lily (mxl):
-       beats = mxl.get_maybe_exist_named_child ('beats')
-       type = mxl.get_maybe_exist_named_child ('beat-type')
-       change = musicexp.TimeSignatureChange()
-       change.fraction = (string.atoi(beats.get_text ()),
-                          string.atoi(type.get_text ()))
-       
-       return change
-
-def musicxml_key_to_lily (mxl):
-       start_pitch  = musicexp.Pitch ()
-       try:
-               mode = mxl.get_maybe_exist_named_child ('mode')
-               if mode:
-                       mode = mode.get_text ()
-               else:
-                       mode = 'major'
-                       
-               (n,a) = { 'major' : (0,0),
-                         'minor' : (6,0),
-                       }[mode]
-               start_pitch.step = n
-               start_pitch.alteration = a
-       except  KeyError:
-               print 'unknown mode', mode
-               
-       fifths = string.atoi (mxl.get_maybe_exist_named_child ('fifths').get_text ())
-
-       fifth = musicexp.Pitch()
-       fifth.step = 4
-       if fifths < 0:
-               fifths *= -1
-               fifth.step *= -1
-               fifth.normalize ()
-       
-       start_pitch = musicexp.Pitch()
-       for x in range (fifths):
-               start_pitch = start_pitch.transposed (fifth)
-
-       start_pitch.octave = 0
-
-       change = musicexp.KeySignatureChange()
-       change.mode = mode
-       change.tonic = start_pitch
-       return change
-       
+    """Collect Musics from
+    MUSIC_LIST demarcated by EVENTS_LIST in TimeScaledMusic objects.
+    """
+
+    
+    indices = []
+
+    j = 0
+    for (ev_chord, tuplet_elt, fraction) in events:
+        while (j < len (music_list)):
+            if music_list[j]== ev_chord:
+                break
+            j += 1
+        if tuplet_elt.type == 'start':
+            indices.append ((j, None, fraction))
+        elif tuplet_elt.type == 'stop':
+            indices[-1] = (indices[-1][0], j, indices[-1][2])
+
+    new_list = []
+    last = 0
+    for (i1, i2, frac) in indices:
+        if i1 >= i2:
+            continue
+
+        new_list.extend (music_list[last:i1])
+        seq = musicexp.SequentialMusic ()
+        last = i2 + 1
+        seq.elements = music_list[i1:last]
+
+        tsm = musicexp.TimeScaledMusic ()
+        tsm.element = seq
+
+        tsm.numerator = frac[0]
+        tsm.denominator  = frac[1]
+
+        new_list.append (tsm)
+
+    new_list.extend (music_list[last:])
+    return new_list
+
+
+def musicxml_clef_to_lily (attributes):
+    change = musicexp.ClefChange ()
+    change.type = attributes.get_clef_sign ()
+    return change
+    
+def musicxml_time_to_lily (attributes):
+    (beats, type) = attributes.get_time_signature ()
+
+    change = musicexp.TimeSignatureChange()
+    change.fraction = (beats, type)
+    
+    return change
+
+def musicxml_key_to_lily (attributes):
+    start_pitch  = musicexp.Pitch ()
+    (fifths, mode) = attributes.get_key_signature () 
+    try:
+        (n,a) = {
+            'major' : (0,0),
+            'minor' : (6,0),
+            }[mode]
+        start_pitch.step = n
+        start_pitch.alteration = a
+    except  KeyError:
+        print 'unknown mode', mode
+
+    fifth = musicexp.Pitch()
+    fifth.step = 4
+    if fifths < 0:
+        fifths *= -1
+        fifth.step *= -1
+        fifth.normalize ()
+    
+    start_pitch = musicexp.Pitch()
+    for x in range (fifths):
+        start_pitch = start_pitch.transposed (fifth)
+
+    start_pitch.octave = 0
+
+    change = musicexp.KeySignatureChange()
+    change.mode = mode
+    change.tonic = start_pitch
+    return change
+    
 def musicxml_attributes_to_lily (attrs):
-       elts = []
-       attr_dispatch =  {
-               'clef': musicxml_clef_to_lily,
-               'time': musicxml_time_to_lily,
-               'key': musicxml_key_to_lily
-       }
-       for (k, func) in attr_dispatch.items ():
-               childs = attrs.get_named_children (k)
-
-               ## ugh: you get clefs spread over staves for piano
-               if childs:
-                       elts.append (func (childs[0]))
-       
-       return elts
-
-def create_skip_music (duration):
-       skip = musicexp.SkipEvent()
-       skip.duration.duration_log = 0
-       skip.duration.factor = duration
-
-       evc = musicexp.EventChord ()
-       evc.append (skip)
-       return evc
+    elts = []
+    attr_dispatch =  {
+        'clef': musicxml_clef_to_lily,
+        'time': musicxml_time_to_lily,
+        'key': musicxml_key_to_lily
+    }
+    for (k, func) in attr_dispatch.items ():
+        childs = attrs.get_named_children (k)
+
+        ## ugh: you get clefs spread over staves for piano
+        if childs:
+            elts.append (func (attrs))
+    
+    return elts
 
 spanner_event_dict = {
-       'slur' : musicexp.SlurEvent,
-       'beam' : musicexp.BeamEvent,
-}      
+    'slur' : musicexp.SlurEvent,
+    'beam' : musicexp.BeamEvent,
+}        
 spanner_type_dict = {
-       'start': -1,
-       'begin': -1,
-       'stop': 1,
-       'end' : 1
+    'start': -1,
+    'begin': -1,
+    'stop': 1,
+    'end' : 1
 }
 
 def musicxml_spanner_to_lily_event (mxl_event):
-       ev = None
-       
-       name = mxl_event.get_name()
-       try:
-               func = spanner_event_dict[name]
-               ev = func()
-       except KeyError:
-               print 'unknown span event ', mxl_event
-
-       try:
-               key = mxl_event.get_type ()
-               ev.span_direction = spanner_type_dict[key]
-       except KeyError:
-               print 'unknown span type', key, 'for', name
-
-       return ev
+    ev = None
+    
+    name = mxl_event.get_name()
+    try:
+        func = spanner_event_dict[name]
+        ev = func()
+    except KeyError:
+        print 'unknown span event ', mxl_event
+
+    try:
+        key = mxl_event.get_type ()
+        ev.span_direction = spanner_type_dict[key]
+    except KeyError:
+        print 'unknown span type', key, 'for', name
+
+    return ev
+
+instrument_drumtype_dict = {
+    'Acoustic Snare Drum': 'acousticsnare',
+    'Side Stick': 'sidestick',
+    'Open Triangle': 'opentriangle',
+    'Mute Triangle': 'mutetriangle',
+    'Tambourine': 'tambourine',
+    
+}
 
 def musicxml_note_to_lily_main_event (n):
-       pitch  = None
-       duration = None
-               
-       mxl_pitch = n.get_maybe_exist_typed_child (musicxml.Pitch)
-       event = None
-       if mxl_pitch:
-               pitch = musicxml_pitch_to_lily (mxl_pitch)
-               event = musicexp.NoteEvent()
-               event.pitch = pitch
-
-               acc = n.get_maybe_exist_named_child ('accidental')
-               if acc:
-                       # let's not force accs everywhere. 
-                       event.cautionary = acc.editorial
-               
-       elif n.get_maybe_exist_typed_child (musicxml.Rest):
-               event = musicexp.RestEvent()
-
-       event.duration = musicxml_duration_to_lily (n)
-       return event
-
+    pitch  = None
+    duration = None
+        
+    mxl_pitch = n.get_maybe_exist_typed_child (musicxml.Pitch)
+    event = None
+    if mxl_pitch:
+        pitch = musicxml_pitch_to_lily (mxl_pitch)
+        event = musicexp.NoteEvent()
+        event.pitch = pitch
+
+        acc = n.get_maybe_exist_named_child ('accidental')
+        if acc:
+            # let's not force accs everywhere. 
+            event.cautionary = acc.editorial
+        
+    elif n.get_maybe_exist_typed_child (musicxml.Rest):
+        event = musicexp.RestEvent()
+    elif n.instrument_name:
+        event = musicexp.NoteEvent ()
+        event.drum_type = instrument_drumtype_dict[n.instrument_name]
+        
+    
+    if not event:
+        n.message ("could not find suitable event")
+
+    event.duration = musicxml_duration_to_lily (n)
+    return event
+
+
+## todo
+class NegativeSkip:
+    def __init__ (self, here, dest):
+        self.here = here
+        self.dest = dest
+
+class LilyPondVoiceBuilder:
+    def __init__ (self):
+        self.elements = []
+        self.end_moment = Rational (0)
+        self.begin_moment = Rational (0)
+        self.pending_multibar = Rational (0)
+
+    def _insert_multibar (self):
+        r = musicexp.MultiMeasureRest ()
+        r.duration = musicexp.Duration()
+        r.duration.duration_log = 0
+        r.duration.factor = self.pending_multibar
+        self.elements.append (r)
+        self.begin_moment = self.end_moment
+        self.end_moment = self.begin_moment + self.pending_multibar
+        self.pending_multibar = Rational (0)
+        
+    def add_multibar_rest (self, duration):
+        self.pending_multibar += duration
+        
+        
+    def add_music (self, music, duration):
+        assert isinstance (music, musicexp.Music)
+        if self.pending_multibar > Rational (0):
+            self._insert_multibar ()
+
+        self.elements.append (music)
+        self.begin_moment = self.end_moment
+        self.end_moment = self.begin_moment + duration 
+
+    def add_bar_check (self, number):
+        b = musicexp.BarCheck ()
+        b.bar_number = number
+        self.add_music (b, Rational (0))
+
+    def jumpto (self, moment):
+        current_end = self.end_moment + self.pending_multibar
+        diff = moment - current_end
+        
+        if diff < Rational (0):
+            raise NegativeSkip(current_end, moment)
+
+        if diff > Rational (0):
+            skip = musicexp.SkipEvent()
+            skip.duration.duration_log = 0
+            skip.duration.factor = diff
+
+            evc = musicexp.EventChord ()
+            evc.elements.append (skip)
+            self.add_music (evc, diff)
+                
+    def last_event_chord (self, starting_at):
+
+        value = None
+        if (self.elements
+            and isinstance (self.elements[-1], musicexp.EventChord)
+            and self.begin_moment == starting_at):
+            value = self.elements[-1]
+        else:
+            self.jumpto (starting_at)
+            value = None
+
+        return value
+        
+    def correct_negative_skip (self, goto):
+        self.end_moment = goto
+        self.begin_moment = goto
+        evc = musicexp.EventChord ()
+        self.elements.append (evc)
+        
 def musicxml_voice_to_lily_voice (voice):
-       ly_voice = []
-       ly_now = Rational (0)
-       pending_skip = Rational (0) 
-
-       tuplet_events = []
-
-       for n in voice._elements:
-               if n.get_name () == 'forward':
-                       continue
-               
-               if isinstance (n, musicxml.Attributes):
-                       ly_now += pending_skip
-                       pending_skip = Rational (0)
-                       
-                       ly_voice.extend (musicxml_attributes_to_lily (n))
-                       continue
-               
-               if not n.__class__.__name__ == 'Note':
-                       print 'not a Note or Attributes?', n
-                       continue
-
-               if n.is_first () and ly_voice:
-                       ly_voice[-1].comment += '\n'
-               
-
-               main_event = musicxml_note_to_lily_main_event (n)
-
-               ev_chord = None
-               if None ==  n.get_maybe_exist_typed_child (musicxml.Chord):
-                       ly_now += pending_skip
-                       pending_skip = main_event.get_length ()
-
-                       if ly_now <> n._when:
-                               diff = n._when - ly_now
-
-                               if diff < Rational (0):
-                                       print 'huh: negative skip', n._when, ly_now, n._duration
-                                       diff = Rational (1,314159265)
-
-                               ly_voice.append (create_skip_music (diff))
-                               ly_now = n._when
-                               
-                       ly_voice.append (musicexp.EventChord())
-               else:
-                       pass
-               
-               ev_chord = ly_voice[-1]
-               ev_chord.append (main_event)
-               
-               notations = n.get_maybe_exist_typed_child (musicxml.Notations)
-               tuplet_event = None
-               span_events = []
-               if notations:
-                       if notations.get_tuplet():
-                               tuplet_event = notations.get_tuplet()
-                               mod = n.get_maybe_exist_typed_child (musicxml.Time_modification)
-                               frac = (1,1)
-                               if mod:
-                                       frac = mod.get_fraction ()
-                               
-                               tuplet_events.append ((ev_chord, tuplet_event, frac))
-
-                       slurs = [s for s in notations.get_named_children ('slur')
-                                if s.get_type () in ('start','stop')]
-                       if slurs:
-                               if len (slurs) > 1:
-                                       print 'more than 1 slur?'
-
-                               lily_ev = musicxml_spanner_to_lily_event (slurs[0])
-                               ev_chord.append (lily_ev)
-
-                       mxl_tie = notations.get_tie ()
-                       if mxl_tie and mxl_tie.type == 'start':
-                               ev_chord.append (musicexp.TieEvent ())
-
-               mxl_beams = [b for b in n.get_named_children ('beam')
-                            if b.get_type () in ('begin', 'end')] 
-               if mxl_beams:
-                       beam_ev = musicxml_spanner_to_lily_event (mxl_beams[0])
-                       if beam_ev:
-                               ev_chord.append (beam_ev)
-                       
-               if tuplet_event:
-                       mod = n.get_maybe_exist_typed_child (musicxml.Time_modification)
-                       frac = (1,1)
-                       if mod:
-                               frac = mod.get_fraction ()
-                               
-                       tuplet_events.append ((ev_chord, tuplet_event, frac))
-
-       ly_voice = group_tuplets (ly_voice, tuplet_events)
-
-       seq_music = musicexp.SequentialMusic()
-       
-       seq_music.elements = ly_voice
-       return seq_music
+    tuplet_events = []
+    modes_found = {}
+
+    voice_builder = LilyPondVoiceBuilder()
+
+    for n in voice._elements:
+        if n.get_name () == 'forward':
+            continue
+
+        if not n.get_maybe_exist_named_child ('chord'):
+            try:
+                voice_builder.jumpto (n._when)
+            except NegativeSkip, neg:
+                voice_builder.correct_negative_skip (n._when)
+                n.message ("Negative skip? from %s to %s, diff %s" % (neg.here, neg.dest, neg.dest - neg.here))
+            
+        if isinstance (n, musicxml.Attributes):
+            if n.is_first () and n._measure_position == Rational (0):
+                voice_builder.add_bar_check (int (n.get_parent ().number))
+            for a in musicxml_attributes_to_lily (n):
+                voice_builder.add_music (a, Rational (0))
+            continue
+
+        if not n.__class__.__name__ == 'Note':
+            print 'not a Note or Attributes?', n
+            continue
+
+        rest = n.get_maybe_exist_typed_child (musicxml.Rest)
+        if (rest
+            and rest.is_whole_measure ()):
+
+            voice_builder.add_multibar_rest (n._duration)
+            continue
+
+        if n.is_first () and n._measure_position == Rational (0):
+            num = int (n.get_parent ().number)
+            voice_builder.add_bar_check (num)
+        
+        main_event = musicxml_note_to_lily_main_event (n)
+
+        try:
+            if main_event.drum_type:
+                modes_found['drummode'] = True
+        except AttributeError:
+            pass
+
+
+        ev_chord = voice_builder.last_event_chord (n._when)
+        if not ev_chord: 
+            ev_chord = musicexp.EventChord()
+            voice_builder.add_music (ev_chord, n._duration)
+
+        ev_chord.append (main_event)
+        
+        notations = n.get_maybe_exist_typed_child (musicxml.Notations)
+        tuplet_event = None
+        span_events = []
+        if notations:
+            if notations.get_tuplet():
+                tuplet_event = notations.get_tuplet()
+                mod = n.get_maybe_exist_typed_child (musicxml.Time_modification)
+                frac = (1,1)
+                if mod:
+                    frac = mod.get_fraction ()
+                
+                tuplet_events.append ((ev_chord, tuplet_event, frac))
+
+            slurs = [s for s in notations.get_named_children ('slur')
+                if s.get_type () in ('start','stop')]
+            if slurs:
+                if len (slurs) > 1:
+                    print 'more than 1 slur?'
+
+                lily_ev = musicxml_spanner_to_lily_event (slurs[0])
+                ev_chord.append (lily_ev)
+
+            mxl_tie = notations.get_tie ()
+            if mxl_tie and mxl_tie.type == 'start':
+                ev_chord.append (musicexp.TieEvent ())
+
+        mxl_beams = [b for b in n.get_named_children ('beam')
+                     if (b.get_type () in ('begin', 'end')
+                         and b.is_primary ())] 
+        if mxl_beams:
+            beam_ev = musicxml_spanner_to_lily_event (mxl_beams[0])
+            if beam_ev:
+                ev_chord.append (beam_ev)
+            
+        if tuplet_event:
+            mod = n.get_maybe_exist_typed_child (musicxml.Time_modification)
+            frac = (1,1)
+            if mod:
+                frac = mod.get_fraction ()
+                
+            tuplet_events.append ((ev_chord, tuplet_event, frac))
+
+    ## force trailing mm rests to be written out.   
+    voice_builder.add_music (musicexp.EventChord (), Rational (0))
+    
+    ly_voice = group_tuplets (voice_builder.elements, tuplet_events)
 
+    seq_music = musicexp.SequentialMusic()
 
-def musicxml_id_to_lily (id):
-       digits = ['one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight',
-                 'nine', 'ten']
-       
-       for dig in digits:
-               d = digits.index (dig) + 1
-               dig = dig[0].upper() + dig[1:]
-               id = re.sub ('%d' % d, dig, id)
+    if 'drummode' in modes_found.keys ():
+        ## \key <pitch> barfs in drummode.
+        ly_voice = [e for e in ly_voice
+                    if not isinstance(e, musicexp.KeySignatureChange)]
+    
+    seq_music.elements = ly_voice
 
-       id = re.sub  ('[^a-zA-Z]', 'X', id)
-       return id
+    
+    
+    if len (modes_found) > 1:
+       print 'Too many modes found', modes_found.keys ()
+
+    return_value = seq_music
+    for mode in modes_found.keys ():
+        v = musicexp.ModeChangingMusicWrapper()
+        v.element = return_value
+        v.mode = mode
+        return_value = v
+    
+    return return_value
 
 
-def musicxml_pitch_to_lily (mxl_pitch):
-       p = musicexp.Pitch()
-       p.alteration = mxl_pitch.get_alteration ()
-       p.step = (ord (mxl_pitch.get_step ()) - ord ('A') + 7 - 2) % 7
-       p.octave = mxl_pitch.get_octave () - 4
-       return p
+def musicxml_id_to_lily (id):
+    digits = ['one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight',
+              'nine', 'ten']
+    
+    for dig in digits:
+        d = digits.index (dig) + 1
+        dig = dig[0].upper() + dig[1:]
+        id = re.sub ('%d' % d, dig, id)
 
+    id = re.sub  ('[^a-zA-Z]', 'X', id)
+    return id
 
 
+def musicxml_pitch_to_lily (mxl_pitch):
+    p = musicexp.Pitch()
+    p.alteration = mxl_pitch.get_alteration ()
+    p.step = (ord (mxl_pitch.get_step ()) - ord ('A') + 7 - 2) % 7
+    p.octave = mxl_pitch.get_octave () - 4
+    return p
+
 def voices_in_part (part):
-       """Return a Name -> Voice dictionary for PART"""
-       part.interpret ()
-       part.extract_voices ()          
-       voice_dict = part.get_voices ()
+    """Return a Name -> Voice dictionary for PART"""
+    part.interpret ()
+    part.extract_voices ()
+    voice_dict = part.get_voices ()
 
-       return voice_dict
+    return voice_dict
 
 def voices_in_part_in_parts (parts):
-       """return a Part -> Name -> Voice dictionary"""
-       return dict([(p, voices_in_part (p)) for p in parts])
+    """return a Part -> Name -> Voice dictionary"""
+    return dict([(p, voices_in_part (p)) for p in parts])
 
 
 def get_all_voices (parts):
-       all_voices = voices_in_part_in_parts (parts)
-
-       all_ly_voices = {}
-       for p, name_voice in all_voices.items ():
-
-               part_ly_voices = {}
-               for n, v in name_voice.items ():
-                       progress ("Converting to LilyPond expressions...")
-                       part_ly_voices[n] = (musicxml_voice_to_lily_voice (v), v)
-
-               all_ly_voices[p] = part_ly_voices
-               
-       return all_ly_voices
-
-class NonDentedHeadingFormatter (optparse.IndentedHelpFormatter):
-    def format_heading(self, heading):
-           if heading:
-                   return heading[0].upper() + heading[1:] + ':\n'
-           return ''
-    def format_option_strings(self, option):
-           sep = ' '
-           if option._short_opts and option._long_opts:
-                   sep = ','
-
-           metavar = ''
-           if option.takes_value():
-                   metavar = '=' + option.metavar or option.dest.upper()
-
-           return "%3s%s %s%s" % (" ".join (option._short_opts),
-                                  sep,
-                                  " ".join (option._long_opts),
-                                  metavar)
-
-    def format_usage(self, usage):
-        return _("Usage: %s\n") % usage
-    
-    def format_description(self, description):
-           return description
-  
+    all_voices = voices_in_part_in_parts (parts)
+
+    all_ly_voices = {}
+    for p, name_voice in all_voices.items ():
+
+        part_ly_voices = {}
+        for n, v in name_voice.items ():
+            progress ("Converting to LilyPond expressions...")
+            part_ly_voices[n] = (musicxml_voice_to_lily_voice (v), v)
+
+        all_ly_voices[p] = part_ly_voices
+        
+    return all_ly_voices
+
 
 def option_parser ():
-       p = optparse.OptionParser(usage='musicxml2ly FILE.xml',
-                                 version = """%prog (LilyPond) @TOPLEVEL_VERSION@
+    p = ly.get_option_parser(usage='musicxml2ly FILE.xml',
+                 version = """%prog (LilyPond) @TOPLEVEL_VERSION@
 
 This program is free software.  It is covered by the GNU General Public
 License and you are welcome to change it and/or distribute copies of it
@@ -404,146 +491,181 @@ under certain conditions.  Invoke as `lilypond --warranty' for more
 information.
 
 Copyright (c) 2005--2006 by
-  Han-Wen Nienhuys <hanwen@xs4all.nl> and
-  Jan Nieuwenhuizen <janneke@gnu.org>
+    Han-Wen Nienhuys <hanwen@xs4all.nl> and
+    Jan Nieuwenhuizen <janneke@gnu.org>
 """,
 
-                                 description  =
-                                 """Convert MusicXML file to LilyPond input.
+                 description  =
+                 """Convert MusicXML file to LilyPond input.
 """
-                                 )
-       p.add_option ('-v', '--verbose',
-                     action = "store_true",
-                     dest = 'verbose',
-                     help = 'be verbose')
-       p.add_option ('-o', '--output',
-                     metavar = 'FILE',
-                     action = "store",
-                     default = None,
-                     type = 'string',
-                     dest = 'output',
-                     help = 'set output file')
-
-       p.add_option_group  ('', description = '''Report bugs via http://post.gmane.org/post.php?group=gmane.comp.gnu.lilypond.bugs
+                 )
+    p.add_option ('-v', '--verbose',
+                  action = "store_true",
+                  dest = 'verbose',
+                  help = 'be verbose')
+
+    p.add_option ('', '--lxml',
+                  action="store_true",
+                  default=False,
+                  dest="use_lxml",
+                  help="Use lxml.etree; uses less memory and cpu time.")
+    
+    p.add_option ('-o', '--output',
+                  metavar = 'FILE',
+                  action="store",
+                  default=None,
+                  type='string',
+                  dest='output_name',
+                  help='set output file')
+
+    p.add_option_group  ('', description = '''Report bugs via http://post.gmane.org/post.php?group=gmane.comp.gnu.lilypond.bugs
 ''')
-       p.formatter = NonDentedHeadingFormatter () 
-       return p
+    return p
 
 def music_xml_voice_name_to_lily_name (part, name):
-       str = "Part%sVoice%s" % (part.id, name)
-       return musicxml_id_to_lily (str) 
+    str = "Part%sVoice%s" % (part.id, name)
+    return musicxml_id_to_lily (str) 
 
 def print_voice_definitions (printer, voices):
-       for (part, nv_dict) in voices.items():
-               for (name, (voice, mxlvoice)) in nv_dict.items ():
-                       k = music_xml_voice_name_to_lily_name (part, name)
-                       printer.dump ('%s = ' % k)
-                       voice.print_ly (printer)
-                       printer.newline()
+    for (part, nv_dict) in voices.items():
+        
+        for (name, (voice, mxlvoice)) in nv_dict.items ():
+            k = music_xml_voice_name_to_lily_name (part, name)
+            printer.dump ('%s = ' % k)
+            voice.print_ly (printer)
+            printer.newline()
+
+            
 def uniq_list (l):
-       return dict ([(elt,1) for elt in l]).keys ()
-       
+    return dict ([(elt,1) for elt in l]).keys ()
+    
 def print_score_setup (printer, part_list, voices):
-       part_dict = dict ([(p.id, p) for p in voices.keys ()]) 
-
-       printer ('<<')
-       printer.newline ()
-       for part_definition in part_list:
-               part_name = part_definition.id
-               try:
-                       part = part_dict[part_name]
-               except KeyError:
-                       print 'unknown part in part-list:', part_name
-                       continue
-
-               nv_dict = voices[part]
-               staves = reduce (lambda x,y: x+ y,
-                                [mxlvoice._staves.keys ()
-                                 for (v, mxlvoice) in nv_dict.values ()],
-                                [])
-
-               if len (staves) > 1:
-                       staves = uniq_list (staves)
-                       staves.sort ()
-                       printer ('\\context PianoStaff << ')
-                       printer.newline ()
-                       
-                       for s in staves:
-                               staff_voices = [music_xml_voice_name_to_lily_name (part, voice_name)
-                                               for (voice_name, (v, mxlvoice)) in nv_dict.items ()
-                                               if mxlvoice._start_staff == s]
-                               
-                               printer ('\\context Staff = "%s" << ' % s)
-                               printer.newline ()
-                               for v in staff_voices:
-                                       printer ('\\context Voice = "%s"  \\%s' % (v,v))
-                                       printer.newline ()
-                               printer ('>>')
-                               printer.newline ()
-                               
-                       printer ('>>')
-                       printer.newline ()
-                       
-               else:
-                       printer ('\\new Staff <<')
-                       printer.newline ()
-                       for (n,v) in nv_dict.items ():
-
-                               n = music_xml_voice_name_to_lily_name (part, n) 
-                               printer ('\\context Voice = "%s"  \\%s' % (n,n))
-                       printer ('>>')
-                       printer.newline ()
-                       
-
-       printer ('>>')
-       printer.newline ()
-
-                               
+    part_dict = dict ([(p.id, p) for p in voices.keys ()]) 
+
+    printer ('<<')
+    printer.newline ()
+    for part_definition in part_list:
+        part_name = part_definition.id
+        try:
+            part = part_dict[part_name]
+        except KeyError:
+            print 'unknown part in part-list:', part_name
+            continue
+
+        nv_dict = voices[part]
+        staves = reduce (lambda x,y: x+ y,
+                [mxlvoice._staves.keys ()
+                 for (v, mxlvoice) in nv_dict.values ()],
+                [])
+
+        if len (staves) > 1:
+            staves = uniq_list (staves)
+            staves.sort ()
+            printer ('\\context PianoStaff << ')
+            printer.newline ()
+            
+            for s in staves:
+                staff_voices = [music_xml_voice_name_to_lily_name (part, voice_name)
+                        for (voice_name, (v, mxlvoice)) in nv_dict.items ()
+                        if mxlvoice._start_staff == s]
+                
+                printer ('\\context Staff = "%s" << ' % s)
+                printer.newline ()
+                for v in staff_voices:
+                    printer ('\\context Voice = "%s"  \\%s' % (v,v))
+                    printer.newline ()
+                printer ('>>')
+                printer.newline ()
+                
+            printer ('>>')
+            printer.newline ()
+            
+        else:
+            printer ('\\new Staff <<')
+            printer.newline ()
+            for (n,v) in nv_dict.items ():
+
+                n = music_xml_voice_name_to_lily_name (part, n) 
+                printer ('\\context Voice = "%s"  \\%s' % (n,n))
+            printer ('>>')
+            printer.newline ()
+            
+    printer ('>>')
+    printer.newline ()
 
 def print_ly_preamble (printer, filename):
-       printer.dump_version ()
-       printer.print_verbatim ('%% converted from %s\n' % filename)
-
-def convert (filename, output_name):
-       printer = musicexp.Output_printer()
-       progress ("Reading MusicXML...")
-       
-       tree = musicxml.read_musicxml (filename)
-       parts = tree.get_typed_children (musicxml.Part)
-       voices = get_all_voices (parts)
-
-       part_list = []
-       if tree.get_maybe_exist_typed_child (musicxml.Part_list):
-               pl = tree.get_maybe_exist_typed_child (musicxml.Part_list)
-               part_list = pl.get_named_children ("score-part")
-               
-       if not output_name:
-               output_name = os.path.basename (filename)
-               output_name = os.path.splitext (output_name)[0] + '.ly'
-
-               
-       if output_name:
-               progress ("Output to `%s'" % output_name)
-               printer.set_file (open (output_name, 'w'))
-       
-       progress ("Printing as .ly...")
-
-       print_ly_preamble (printer, filename)
-       print_voice_definitions (printer,  voices)
-       print_score_setup (printer, part_list, voices)
-       printer.newline ()
-       return voices
+    printer.dump_version ()
+    printer.print_verbatim ('%% converted from %s\n' % filename)
+
+def read_musicxml (filename, use_lxml):
+    if use_lxml:
+        import lxml.etree
+        
+        tree = lxml.etree.parse (filename)
+        mxl_tree = musicxml.lxml_demarshal_node (tree.getroot ())
+        return mxl_tree
+    else:
+        from xml.dom import minidom, Node
+        
+        doc = minidom.parse(filename)
+        node = doc.documentElement
+        return musicxml.minidom_demarshal_node (node)
+
+    return None
+
+
+def convert (filename, options):
+    progress ("Reading MusicXML from %s ..." % filename)
+    
+    tree = read_musicxml (filename, options.use_lxml)
+
+    part_list = []
+    id_instrument_map = {}
+    if tree.get_maybe_exist_typed_child (musicxml.Part_list):
+        mxl_pl = tree.get_maybe_exist_typed_child (musicxml.Part_list)
+        part_list = mxl_pl.get_named_children ("score-part")
+        
+    parts = tree.get_typed_children (musicxml.Part)
+    voices = get_all_voices (parts)
+
+    if not options.output_name:
+        options.output_name = os.path.basename (filename) 
+        options.output_name = os.path.splitext (options.output_name)[0]
+
+
+    defs_ly_name = options.output_name + '-defs.ly'
+    driver_ly_name = options.output_name + '.ly'
+
+    printer = musicexp.Output_printer()
+    progress ("Output to `%s'" % defs_ly_name)
+    printer.set_file (open (defs_ly_name, 'w'))
+
+    print_ly_preamble (printer, filename)
+    print_voice_definitions (printer, voices)
+    
+    printer.close ()
+    
+    
+    progress ("Output to `%s'" % driver_ly_name)
+    printer = musicexp.Output_printer()
+    printer.set_file (open (driver_ly_name, 'w'))
+    print_ly_preamble (printer, filename)
+    printer.dump (r'\include "%s"' % defs_ly_name)
+    print_score_setup (printer, part_list, voices)
+    printer.newline ()
+
+    return voices
 
 
 def main ():
-       opt_parser = option_parser()
+    opt_parser = option_parser()
 
-       (options, args) = opt_parser.parse_args ()
-       if not args:
-               opt_parser.print_usage()
-               sys.exit (2)
+    (options, args) = opt_parser.parse_args ()
+    if not args:
+        opt_parser.print_usage()
+        sys.exit (2)
 
-       voices = convert (args[0], options.output)
+    voices = convert (args[0], options)
 
 if __name__ == '__main__':
-       main()
+    main()