openmedialibrary/oml/meta/__init__.py
2014-05-17 17:16:18 +02:00

62 lines
1.8 KiB
Python

# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4
from __future__ import division
import logging
logger = logging.getLogger('meta')
import abebooks
import loc
import lookupbyisbn
import openlibrary
import worldcat
import google
import duckduckgo
providers = [
('openlibrary', 'olid'),
('loc', 'lccn'),
('worldcat', 'oclc'),
('lookupbyisbn', 'asin'),
('abebooks', 'isbn10')
]
def find(title, author=None, publisher=None, date=None):
#results = google.find(title=title, author=author, publisher=publisher, date=date)
results = duckduckgo.find(title=title, author=author, publisher=publisher, date=date)
'''
results = openlibrary.find(title=title, author=author, publisher=publisher, date=date)
for r in results:
r['mainid'] = 'olid'
'''
return results
def lookup(key, value):
data = {key: value}
ids = [(key, value)]
provider_data = {}
done = False
while not done:
done = True
for provider, id in providers:
for key, value in ids:
for kv in globals()[provider].get_ids(key, value):
if not kv in ids:
ids.append(kv)
done = False
logger.debug('lookup %s=%s => %s', ids[0][0], ids[0][1], ids)
for k, v in ids:
for provider, id in providers:
if id == k and provider not in provider_data:
provider_data[provider] = globals()[provider].lookup(v)
for provider in sorted(
provider_data.keys(),
key=lambda x: -len(provider_data[x])
):
logger.debug('%s %s %s', provider, len(provider_data[provider]), provider_data[provider].keys())
for k_, v_ in provider_data[provider].iteritems():
if not k_ in data:
data[k_] = v_
return data