baidu_mp3_album_downloader.py

#!/usr/bin/env python2
import re, sys, os, urllib

re_songs = re.compile(r'\'sid\': \'(.+?)\', \'sname\': \'(.+?)\'')
re_name = re.compile(r'\\&.+?;')
template_wgets = 'wget -nv -c -O "%s" %s'
durl = 'http://yinyueyun.baidu.com/data/cloud/downloadsongfile\?songIds\=%s\&rate\=320'

def modified_sname(sname):
    sname = sname.replace('\\'', '\'')
    sname = sname.replace(''', '\'')
    sname = sname.replace('\\’', '\'')
    sname = sname.replace('’', '\'')
    sname = re_name.sub(r' - ', sname)      # remove all http symbol from song-name.
    sname = sname.replace('/', ' - ')
    if len(sname) >= 250:
        return sname[:243] + '...mp3'
    else:
        return sname

def down_album(albumid):
    f = urllib.urlopen('http://music.baidu.com/album/' + albumid).read()
    infos = re_songs.findall(f)
    size = len(infos)
    z = 0
    if size <= 9:
        z = 1
    elif size >= 10 and size <= 99:
        z = 2
    elif size >= 100 and size <= 999:
        z = 3
    else:
        z = 1

    for i in range(1, size + 1):
        songid, sname = infos[i - 1]
        sname = str(i).zfill(z) + '.' + sname.strip() + '.mp3'
        sname = modified_sname(sname)

        wget = template_wgets % (sname, durl % songid)
        status = os.system(wget)
        if status == 2048:           # 320kbps-mp3 is not available, so getting 192kbps-mp3.
            wget = template_wgets % (sname, (durl[:-3]  + '192') % songid)
            status = os.system(wget)
        if status == 2048:           # 320kbps-mp3 is not available, so getting 128kbps-mp3.
            wget = template_wgets % (sname, (durl[:-3] + '128') % songid)
            status = os.system(wget)
        if status not in [0, 2048]:     # other http-errors, such as 302.
            print '\n\n ----### ERROR ==> %d ###--- \n\n' % status
            break

def down_song(songids):
    for songid in songids:
        f = urllib.urlopen('http://music.baidu.com/song/' + songid).read()
        infos = re.search(r'\'sid\': \'%s\', \'sname\': \'(.+?)\'' % songid, f)
        try:
            sname = infos.group(1)
        except:
            infos = re.search(r'<span class="name">(.+?)</span>', f)
            sname = infos.group(1)
        sname = sname.strip() + '.mp3'
        sname = modified_sname(sname)

        wget = template_wgets % (sname, durl % songid)
        status = os.system(wget)
        if status == 2048:           # 320kbps-mp3 is not available, so getting 192kbps-mp3.
            wget = template_wgets % (sname, (durl[:-3]  + '192') % songid)
            status = os.system(wget)
        if status == 2048:           # 320kbps-mp3 is not available, so getting 128kbps-mp3.
            wget = template_wgets % (sname, (durl[:-3] + '128') % songid)
            status = os.system(wget)
        if status not in [0, 2048]:     # other http-errors, such as 302.
            print '\n\n ----### ERROR ==> %d ###--- \n\n' % status
            break

if __name__ == '__main__':
    argv = sys.argv
    if argv[1] == '-s' and argv[2]:
        down_song(argv[2:])
    elif argv[1] == '-a' and argv[2]:
        down_album(argv[2])
    else:
        print 'Usage:\n  -s songid1 songid2 songidn     download some songs.\n  -a albumid    download an album.'