openmedialibrary/oml/meta/google.py

36 lines
853 B
Python
Raw Normal View History

2014-05-16 08:06:11 +00:00
# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4
2014-09-02 22:32:44 +00:00
2014-05-16 08:06:11 +00:00
import ox.web.google
import stdnum.isbn
from .utils import find_isbns
2014-05-17 14:26:59 +00:00
import logging
logger = logging.getLogger('meta.google')
2014-05-16 08:06:11 +00:00
2014-05-21 00:02:21 +00:00
def find(query):
logger.debug('find %s', query)
2014-05-16 08:06:11 +00:00
query += ' isbn'
isbns = []
for r in ox.web.google.find(query):
isbns += find_isbns(' '.join(r))
2014-05-17 14:26:59 +00:00
logger.debug('isbns', isbns)
2014-05-16 08:06:11 +00:00
results = []
done = set()
for isbn in isbns:
if isbn not in done:
r = {
2014-05-21 00:02:21 +00:00
'isbn': isbn,
'primaryid': ['isbn', isbn]
2014-05-16 08:06:11 +00:00
}
results.append(r)
done.add(isbn)
if len(isbn) == 10:
done.add(stdnum.isbn.to_isbn13(isbn))
2014-05-21 00:02:21 +00:00
if len(isbn) == 13 and isbn.startswith('978'):
2014-05-17 09:19:32 +00:00
done.add(stdnum.isbn.to_isbn10(isbn))
2014-05-16 08:06:11 +00:00
return results