extract cover art from remote mp3

Question

I need to extract a cover art from a remote mp3 file and save it to a file without downloading the whole mp3. But I have no success with it. I have try to download the first 100 bytes of the file like:

import urllib2
from mutagen.mp3 import MP3

req = urllib2.Request('http://www.stephaniequinn.com/Music/Commercial%20DEMO%20-%2001.mp3')
req.headers['Range'] = 'bytes=%s-%s' % (0, 100)
response = urllib2.urlopen(req)
headers = response.info()
print headers.type
print headers.maintype

data = response.read()
print len(data)

I have read about that the id 3 Tags are at the last 128 bytes of a mp3. Now I need some help help to download only the last bytes that contains the apic cover art and extract the image.

thanks for helping me out

the mp3 you link to in your example doesn't have any artwork attached. Can you provide one with artwork? — hansaplast, Jan 19 '17 at 16:05
if I download this mp3 VLC show some cover art... I search for another please wait — fteinz, Jan 19 '17 at 16:13
are you using linux or windows? You'll need some command line tools to find the position of the cover art in the mp3 file — hansaplast, Jan 19 '17 at 16:14
Linux but the script should finally run on Python for Android — fteinz, Jan 19 '17 at 16:16
I also try to handle it with ffmpeg but I don't get it run on Android via Command line — fteinz, Jan 19 '17 at 16:19
How would you extract the cover art? With mutagen? Does that work with partial files? — hansaplast, Jan 19 '17 at 16:34
I hope so... and sorry I still search for a free mp3 direct download — fteinz, Jan 19 '17 at 16:37

lazka · Accepted Answer · 2017-01-19T18:40:20.373

The covert art is at the beginning of the file in an id3v2 tag.

Here is a hacky solution: read until the whole file is read or mutagen doesn't error out. This would read the whole file in case the mp3 isn't an mp3. Ideally you'd pass it a seekable file-like which does buffering, maybe there is a library for that.

# Python 2 or 3
try:
    import urllib2 as request
except ImportError:
    from urllib import request
from io import BytesIO
from mutagen import MutagenError
from mutagen.mp3 import MP3


def get_mp3(url):
    """
    Args:
        url (str)
    Returns:
        mutagen.mp3.MP3
    Raises:
        mutagen.MutagenError
        EnvironmentError
    """

    r = request.urlopen(url)
    try:
        size = 128
        filelike = BytesIO()
        while 1:
            data = r.read(size)
            size *= 2
            filelike.seek(0, 2)
            filelike.write(data)
            filelike.seek(0)
            try:
                return MP3(filelike)
            except MutagenError:
                if not data:
                    raise
                pass
    finally:
        r.close()

try:
    f = get_mp3("http://web.ist.utl.pt/antonio.afonso/www.aadsm.net/libraries/id3/music/Bruno_Walter_-_01_-_Beethoven_Symphony_No_1_Menuetto.mp3")
except (MutagenError, EnvironmentError):
    pass
else:
    if f.tags:
        for frame in f.tags.getall("APIC"):
            print(frame.pprint())

did you check if this does not download the whole file? I assumed that when not providing `Range` in the header it's downloading the whole file despite telling `read` to only load a few bytes? — hansaplast, Jan 19 '17 at 17:59
@hansaplast I don't know. Looking at it using "strace -y -e recvfrom python foo.py" shows that it's at least not loading the whole file. — lazka, Jan 19 '17 at 18:43

score 0 · Answer 2 · edited May 23 '17 at 11:45

ID3 tags are usually in front of the mp3 files, not in the back. I checked some random mp3s from the internet and they had all their ID3 tags in front, although the mp3 format allows them to have it at the end.

If you want to only download the absolute minimum of bytes (as you don't want to waste your mobiles users bandwidth) you'll need to:

do a partial download of 10 bytes and check that the ID3 tags are in front of the file. If not: download the whole file
extract the size from byte 6-9 (note that the bytes have the leftmost bit always set to zero, as described on id3.org)
do a fresh partial download of the size you just computed

After that you'll have downloaded the full ID3 tags and can extract them. Now, mutagen has the limitation that you need to download the first mp3 audio frame as well, otherwise it will throw an exception: mutagen.mp3.HeaderNotFoundError: can't sync to an MPEG frame. If that limitation is ok for you I posted a python solution at a similar question (in fact this is probably a duplicate, I see you also have copied the source code from the question into your question).

If you absolutely want to minimize your download size then you'd probably want to try a few other modules which hopefully don't require to also download the first audio frame.

extract cover art from remote mp3

2 Answers2