Music by year histogram

music_histogram.py

from __future__ import division
import os
from mutagen import File
from mutagen.flac import FLAC, VCFLACDict
from mutagen.mp3 import MP3
from mutagen.apev2 import APEv2
from mutagen.m4a import M4A
from mutagen.oggvorbis import OggVorbis
import matplotlib.pyplot as plt


def get_tracks(start_path = '.'):
    tracks, albums, flacs, cum_length = 0, 0, 0, 0.0
    cum_bitrate = 0.0
    scanned_tracks = 0
    mp3s = 0
    oggs = 0
    flacs = 0
    m4as = 0
    for dirpath, dirnames, filenames in os.walk(start_path):
        if dirnames == []:
            albums += 1
        for f in filenames:
            ext = os.path.splitext(f)[1]
            fullpath = os.path.join(dirpath, f)
            musicextensions = ['.flac', '.mp3', '.mpc', '.wma', '.m4a', '.ogg']
            if any(ext == musicext for musicext in musicextensions):
                tracks += 1
            if ext == '.flac':
                flacs += 1
                track = FLAC(fullpath)
            if ext == '.mp3':
                mp3s += 1
                track = MP3(fullpath)
            if ext == '.m4a':
                m4as += 1
                track = M4A(fullpath)
            if ext == '.ogg':
                oggs += 1
                track = OggVorbis(fullpath)

            length = track.info.length
            cum_length += length
            scanned_tracks += 1
            bitrate = os.path.getsize(fullpath) / length
            cum_bitrate += bitrate

    print cum_length
    print scanned_tracks
    print 8* cum_bitrate / scanned_tracks
    print mp3s+flacs
    print oggs
    return (tracks, albums)

def years(start_path = '.'):
    years = [str(x) for x in range(1960, 2015)]
    histogram = {year: 0 for year in years}
    for dirpath, dirnames, filenames in os.walk(start_path):
        for f in filenames:
            is_track = False
            ext = os.path.splitext(f)[1]
            fullpath = os.path.join(dirpath, f)
            if ext in ['.flac', '.mp3', '.m4a', '.ogg', '.wma']:
                is_track = True
                track = File(fullpath)
            if is_track:
                date = ''
                keys = track.keys()
                if 'date' in keys:
                    date = track.get('date')[0]
                if 'year' in keys:
                    date = track.get('year')[0]
                if 'TDRC' in keys:
                    date = str(track.get('TDRC')[0])
                if '\xa9day' in keys:
                    date = track.get('\xa9day')[0]
                if 'WM/Year' in keys:
                    date = track.get('WM/Year')[0].value
                for year in years:
                    if date.startswith(year):
                        histogram[year] += 1
    return histogram

def plot_histogram(hist):
    ymax = max(hist.values())
    plt.figure()
    x = [int(x) for x in hist.keys()]
    plt.bar(x,hist.values())
    plt.title('Zeitliche Verteilung Musiksammlung') 
    plt.xlabel('Jahr')
    plt.ylabel('Anzahl Songs')
#    plt.axis([0, xmax, 0, ymax])
    plt.show()

if __name__ == '__main__':
    hist = years('/home/psirus/Musik')
    plot_histogram(hist)
    print hist