Investigation of music21 Haydn data

music21.py

import music21
import os
from collections import defaultdict

HAYDN = music21.corpus.getWork('haydn')

def title(h):
    path, mvt = os.path.split(h)
    mvt, ext = os.path.splitext(mvt)
    qrt = os.path.basename(path)
    return qrt, mvt

def parse_path(h):
    """Given a path, returns Opus, Number, Mvmt #"""
    if not 'opus' in h:
        return # fuck the non-quartets

    path, mvt = os.path.split(h)
    mvt, ext = os.path.splitext(mvt)
    qrt = os.path.basename(path)
    # e.g. opus71no2
    # movement2
    mvt = int(mvt[-1])
    qrt = qrt.replace("opus", "")
    if "no" in qrt:
        opus = qrt[:qrt.index('no')]
        number = int(qrt[-1])
        return int(opus), number, mvt
    else:
        opus = int(qrt)
        return opus, mvt

def v(i):
    s = music21.corpus.parse(HAYDN[i])
#    s.show(app=musescore)
    return s.fileFormat

def show_formats():
    for i, path in enumerate(HAYDN):
        format = v(i)
        qrt, mvt = title(path)
        print qrt, mvt, format

def show_incomplete():
    """
    9/3, 33/2 missing 1 mvt each.
    Also missing complete qrts: 
    1/5 (not qrt -- symphony)
    all opus 2
    9/1, 9/4, 9/5, 9/6
    17/4
    """
    counts = defaultdict(int)

    for i, path in enumerate(HAYDN):
        qrt, mvt = title(path)
        counts[qrt] += 1
    for q, i in counts.iteritems():
        if i < 4:
            print q, i


def save_xml(directory='/Users/elaineleisinger/Dropbox/haydn/musicxml'):
    for path in HAYDN:
        parsed = parse_path(path)
        if parsed:
            try:
                quartet, number, mvmt = parsed
                desired_name = '%03d_%dm%d.xml' % (quartet, number, mvmt)
            except ValueError:
                quartet, mvmt = parsed
                desired_name = '%03dm%d.xml' % (quartet, mvmt)
            desired_path = os.path.join(directory, desired_name)

            score = music21.corpus.parse(path)
            tmp_path = score.write()
            os.rename(tmp_path, desired_path)