gistfile1.txt

import json
import urllib
import collections

def get_iterable(x):
    return (x,) if not isinstance(x, (tuple, list)) else x
#     if isinstance(x, collections.Iterable):
#         return x
#     else:
#         return (x,)

def iter_contains_prefix(iter, prefix):
    for s in iter:
        if s.startswith(prefix):
            return True
    return False

def get_url(url):
    f = urllib.urlopen(url)
    c = f.read()
    f.close()
    return c

def get_meta(item_iter):
    i = 0
    for item in item_iter:
        i += 1
        if i % 100 == 0:
            print i
        j = get_url("http://archive.org/metadata/" + item)
        yield json.loads(j);

def item_iterator():
    filename = 'items'
    f = open("items")
    for line in f:
      yield line.strip()

ies = item_iterator()
ems = get_meta(ies)

def file_has_audio(file):
    if (file.get('source') == 'original' and
        file['format'] in ('VBR MP3', 'AIFF', '24bit Flac', 'Flac',
                           'Apple Lossless Audio', 'Advanced Audio Coding')):
        return True;

for md in ems:
    for file in md['files']:
        if file_has_audio(file):
            fexts = file.get('external-identifier', [])
            found_accoustid = iter_contains_prefix(get_iterable(fexts),
                                                   'urn:acoustid')
            if found_accoustid:
                pass
#                 print (file['format'] + ' ' + md['metadata']['identifier'] + 
#                        u"has file " + file['name'].encode('ascii', 'ignore') + u" HAS accoustid")
            else:
                pass
#                 print (file['format'] + ' ' + md['metadata']['identifier'] + 
#                        u"has file " + file['name'].encode('ascii', 'ignore') + u"w/o accoustid")
    
    if found_accoustid:
        mdexts = md.get('external_identifier', [])
        found_mb = iter_contains_prefix(get_iterable(mdexts),
                                        'urn:mb_release')
        if not found_mb:
            print md['metadata']['identifier'] + u' has accoustid but no mb_release'