openmedialibrary/oml/meta/lookupbyisbn.py

65 lines
2.1 KiB
Python
Raw Normal View History

2014-05-14 09:57:11 +00:00
from ox.cache import read_url
2014-05-14 18:46:31 +00:00
from ox import find_re, strip_tags, decode_html
2014-05-14 09:57:11 +00:00
import re
2014-05-14 18:46:31 +00:00
import stdnum.isbn
2014-05-14 09:57:11 +00:00
base = 'http://www.lookupbyisbn.com'
def get_ids(key, value):
ids = []
if key in ('isbn10', 'isbn13', 'asin'):
url = '%s/Search/Book/%s/1' % (base, value)
data = read_url(url).decode('utf-8')
m = re.compile('href="(/Lookup/Book/[^"]+?)"').findall(data)
if m:
asin = m[0].split('/')[-3]
ids.append(('asin', asin))
2014-05-16 08:06:11 +00:00
if key == 'isbn10':
ids.append(('isbn13', stdnum.isbn.to_isbn13(value)))
2014-05-14 18:46:31 +00:00
if key == 'asin':
if stdnum.isbn.is_valid(value):
ids.append(('isbn10', value))
2014-05-14 09:57:11 +00:00
if ids:
print 'lookupbyisbn.get_ids', key, value
print ids
return ids
def lookup(id):
print 'lookupbyisbn.lookup', id
r = {
'asin': id
}
url = '%s/Lookup/Book/%s/%s/1' % (base, id, id)
data = read_url(url).decode('utf-8')
r["title"] = find_re(data, "<h2>(.*?)</h2>")
keys = {
'author': 'Author(s)',
'publisher': 'Publisher',
'date': 'Publication date',
'edition': 'Edition',
'binding': 'Binding',
'volume': 'Volume(s)',
'pages': 'Pages',
}
for key in keys:
r[key] = find_re(data, '<span class="title">%s:</span>(.*?)</li>'% re.escape(keys[key]))
if r[key] == '--':
r[key] = ''
if key == 'pages' and r[key]:
r[key] = int(r[key])
desc = find_re(data, '<h2>Description:<\/h2>(.*?)<div ')
desc = desc.replace('<br /><br />', ' ').replace('<br /> ', ' ').replace('<br />', ' ')
2014-05-16 08:06:11 +00:00
r['description'] = decode_html(strip_tags(desc))
2014-05-14 09:57:11 +00:00
r['cover'] = find_re(data, '<img src="(.*?)" alt="Book cover').replace('._SL160_', '')
2014-05-14 18:46:31 +00:00
for key in r:
if isinstance(r[key], basestring):
r[key] = decode_html(strip_tags(r[key])).strip()
2014-05-16 08:06:11 +00:00
if 'author' in r and isinstance(r['author'], basestring) and r['author']:
2014-05-14 09:57:11 +00:00
r['author'] = [r['author']]
2014-05-16 08:06:11 +00:00
else:
r['author'] = []
if r['description'].lower() == u'Description of this item is not available at this time.'.lower():
r['description'] = ''
2014-05-14 09:57:11 +00:00
return r